laser texturisation of photovoltaic module superstrates...
TRANSCRIPT
Laser texturisation of photovoltaic module
superstrates for enhanced light trapping
performance
A thesis submitted in fulfilment of the requirement for the degree
of
Doctor of Philosophy
David John Moore
(B.Eng)
Supervisors:
Prof. Dermot Brabazon
Prof. Patrick J. McNally
School of Mechanical and Manufacturing Engineering
Dublin City University, Ireland.
September 2016
I
Declaration
I hereby certify that this material, which I now submit for assessment on the
programme of study leading to the award of Ph.D., is entirely my own work, that I
have exercised reasonable care to ensure that the work is original, and does not to the
best of my knowledge breach any law of copyright, and has not been taken from the
work of others save and to the extent that such work has been cited and acknowledged
within the text of my work.
Signed: ___________________________________
ID No.: 54325869
Date: 7th September 2016
II
Dedication
To Enrika
For always being by my side
III
List of Publications
This thesis describes original work that has not previously been submitted for a degree
in Dublin City University or at any other University. The investigations were carried
out in the School of Mechanical & Manufacturing Engineering, Dublin City
University under the supervision of Prof. Dermot Brabazon. This work has been
disseminated through the following publications.
Journal Papers:
Vazquez, M., Moore, D., He, X., Ben Azouz, A., Nesterenko, E., Nexterenko, P.,
Paull, B., Brabazon, D., Focussed ion beam serial sectioning and imaging of
monolithic materials for 3D reconstruction and morphological parameter evaluation,
Analyst, 139, 1, 2014, pp. 99-104
Moore, D., Rahman, M., Dowling, D., McNally, P.J., Brabazon, D., Laser machined
macro and micro structures on glass for enhanced light trapping in solar cells, Applied
Physics A, 110, 3, 2013, pp. 661-665
Moore, D., Krishnamurthy, S., Chao, Y., Wang, Q., Brabazon, D., and McNally, P.J.,
Characteristics of silicon nanocrystals for photovoltaic applications, Physica Status
Solidi A, 1-4, 2010, Vo. 208, Issue 3, pp. 604-607; DOI 10.1002/pssa.201000381
Peer Reviewed Conference Paper:
Moore, D., Brabazon, D., McNally, P.J., Laser texturisation for increased light
absorption is silicon solar cells, Part 2: Proceedings 3rd International Conference on
Sustainable Energy and Environmental Protection, Dublin, Ireland, 12-15 August
2009, pp. 218-224. ISBN: 97809555781-2-0.
IV
Article:
Moore, D., Huet, G., Cousinet, S., Brabazon, D., and McNally, P., Developing a low-
cost, noncontact, 3D optical surface profilometer with surface visualisation and
roughness analysis capabilities in laser-processed surface characterisation, Read-out,
Journal of Instrumentation, Control and Automation, Vol. 11.01, 2011, ISSN 0791-
4369, pp.2-3.
International Conferences, Seminars & Posters:
Moore, D., Rahman, M., Dowling, D., McNally, P.J., Brabazon, D., Laser machined
macro and micro structures on glass for enhanced light trapping in solar cells, Poster
Presentation, COLA Conference, Playa Del Carmen, Mexico, 15th November 2011.
Moore, D., Krishnamurthy, S., Chao, Y., Wang, Q., Brabazon, D., and McNally, P.J.,
Light trapping in alkyl-capped silicon nanocrystals deposited on laser textured glass
superstrates, Oral Presentation, E-MRS 2011 Spring Meeting Symposium R, Nice,
France, 11th May 2011.
Moore, D., Brabazon, D., McNally, P.J., The effects of laser induced glass texturing
on the absorption of light for photovoltaic applications, Oral Presentation, 4th Int.
Conf on Sustainable Energy and Environmental Protection, Politecnico di Bari, Italy
30th June 2010.
Moore, D., Brabazon, D., and McNally, P.J., Pulsed CO2 Laser Induced Micro and
Nanostructures on Quartz Glass, Oral Presentation, E-MRS 2010 Spring Meeting
Symposium R, Strasbourg, France, 11th June 2010.
Moore, D., Krishnamurthy, S., Chao, Y., Wang, Q., Brabazon, D., and McNally, P.J.,
Characteristics of silicon nanocrystals for photovoltaic applications, Poster
Presentation, E-MRS 2010 Spring Meeting Symposium I, Strasbourg, 10th June 2010.
Moore, D., Brabazon, D., McNally, P.J., The effects of laser induced glass texturing
on the absorption of light for photovoltaic applications, Poster Presentation, Faculty
Research Day 2010, Dublin City University, Ireland 12th May 2010.
V
Moore, D., Brabazon, D., McNally, P.J., Laser texturisation of glass superstrate for
increased light absorption in thin-film silicon solar cells, Poster presentation,
‘Innovation fuelling the Smart Society’ IRCSET Symposium 2009, Dublin, Ireland,
September 25th 2009.
VI
Acknowledgements
Firstly, I would like to acknowledge my supervisors, Prof. Dermot Brabazon and Prof.
Patrick McNally. They have demonstrated great encouragement, provided guidance,
assistance, and flexibility, and have shown great patience, all which has allowed me
to take this thesis to where it is today. I have been very fortunate to have the
opportunity to pursue a Ph.D., and for that I am extremely grateful.
I would like to thank all of the technical staff who assisted my along the way through,
namely Michael May, Alan Meehan, Eoin Tuohy, Jim Barry, Cian Merne, Liam
Domican, and the late Martin Johnson.
I would like to extend further thanks to those who helped my learning, namely Alain
Fave of INSA Lyon who was kind enough to host me in his lab at the beginning of
these studies. It was a great experience, and certainly opened my eyes to the world of
semiconductor processing. I would also like all the collaborators who I was lucky
enough to work with during this study.
I would like to give special thanks to my colleagues and friends, Evans, Saba, Tony,
Irina, Ahmed, Sylvain, Arsalan, and Owen, just to name a few. There is no-one else I
would rather have shared this experience with.
I’d like to thank my family, especially my parents for being voices of encouragement
when life gets hard. Your love, support, and hard-work has made me who I am today.
I am eternally grateful to my amazing wife Enrika, without whose support I might
never have had the energy to complete this thesis. A very special thanks goes to my
little man Henry, there is no one else in this world who gives me perspective in the
way that you do. You have changed everything in how I see the world.
Finally, I need to thank the Irish Research Council for funding this research. Without
this funding my Ph.D. study is not something I would have been able to experience.
VII
Table of Contents Declaration ................................................................................................................... I
Dedication ................................................................................................................... II
List of Publications .................................................................................................... III
Acknowledgements .................................................................................................... VI
List of Figures ............................................................................................................ XI
List of Tables........................................................................................................... XVI
Nomenclature .......................................................................................................... XIX
Abstract ........................................................................................................................ 1
Chapter 1 Introduction ............................................................................................ 2
Chapter 2 Literature Review ................................................................................... 7
2.1 Introduction to solar cells .............................................................................. 7
2.2 The mechanics of the generation of photocurrent ......................................... 7
2.3 Silicon as a photovoltaic material ............................................................... 11
2.4 The Air Mass (AM) standard ...................................................................... 11
2.5 Types of wafer based silicon solar cells ...................................................... 13
2.6 Factors limiting the absorption of light in silicon solar cells ...................... 13
2.7 Possible device configurations for effective light trapping ......................... 14
2.8 Important factors in surface texturisation .................................................... 16
2.9 Common front surface textures ................................................................... 16
2.9.1 Pyramidal texture ................................................................................. 16
2.9.2 Random texture .................................................................................... 18
2.9.3 Honeycomb texture .............................................................................. 19
2.9.4 Grooving .............................................................................................. 20
2.10 Laser micromachining in solar cell design .................................................. 22
2.10.1 Design of solar cells on texturised substrates ...................................... 22
2.10.2 Micro-structuring and Nano-structuring of glass substrate with laser . 23
2.11 Computer simulation for optimum texture type .......................................... 24
2.12 Thin-film solar cell designs ......................................................................... 24
2.13 Silicon quantum dot solar cells ................................................................... 25
2.14 Solar Cell superstrate material selection ..................................................... 27
2.14.1 Refractive indices ................................................................................. 27
2.14.2 Laser absorption characteristics ........................................................... 27
2.14.3 Cost and availability ............................................................................. 28
2.14.4 Thermal properties and stability........................................................... 28
VIII
2.14.5 Transparency to AM1.5G spectrum ..................................................... 28
2.15 Literature Review Summary ....................................................................... 29
Chapter 3 Materials & Methods............................................................................ 30
3.1 Design of a Depth-from-Focus based Optical Profilometer ........................ 30
3.1.1 System Specification ............................................................................ 34
3.1.2 Design of the Optical System ............................................................... 35
3.1.3 Design of the Motion Control System ................................................. 36
3.1.4 Image acquisition and processing ........................................................ 37
3.1.5 Image stitching (tiling) ......................................................................... 39
3.1.6 Surface roughness calculation and evaluation ..................................... 39
3.2 Verification of the optical plane of the 3D Profilometer ............................. 42
3.3 Performance analysis of the linear stages of the 3D Optical Profilometer.. 43
3.3.1 Selection of Target Positions for Positional Analysis .......................... 43
3.3.2 Recording of the stage positions using Dial Gauge Indicators ............ 44
3.3.3 Potential sources of measurement uncertainty ..................................... 47
3.4 Verification of 3D measurement performance ............................................ 49
3.4.1 Testing procedure of the 3D measurement performance of the Optical
Profilometer ........................................................................................................ 55
3.4.2 Potential sources of measurement uncertainty in the 3D measurement ...
.............................................................................................................. 60
3.5 CO2 Laser System hardware and specifications .......................................... 61
3.6 Glass materials selection for solar cells superstrates ................................... 62
3.7 Design of Experiments ................................................................................ 62
3.8 ANOVA Analysis ........................................................................................ 65
3.9 Etching of processed glass with Hydrofluoric Acid (HF) ........................... 65
3.10 Integrating sphere setup for reflection / absorption / transmission
measurements ......................................................................................................... 66
3.11 Correcting absorption data to AM1.5G solar irradiance spectrum ............. 67
3.12 SEM and EDXS Analysis ............................................................................ 68
3.13 Channel Dimensional Analysis ................................................................... 69
3.14 Solar cell selection and preparation ............................................................. 70
3.15 Measurements of I-V vs angle of incidence ................................................ 70
3.16 Optical characterisation of alkyl-capped silicon quantum dots ................... 72
3.16.1 Preparation ........................................................................................... 72
3.16.2 Silicon quantum dots size evaluations ................................................. 73
3.16.3 Deposition of silicon quantum dots onto textured glass superstrates .. 75
3.17 Focused Ion Beam Serial Sectioning ........................................................... 76
3.17.1 Chemicals ............................................................................................. 76
IX
3.17.2 Preparation of carbon monolith ............................................................ 76
3.17.3 Image acquisition ................................................................................. 77
3.17.4 Image processing .................................................................................. 79
3.17.5 3D reconstruction and morphological analysis .................................... 81
Chapter 4 Verification of the 3D non-contact optical profilometer...................... 82
4.1 Positional Analysis of the XYZ stage system ............................................. 82
4.1.1 Large scale positional analysis ............................................................. 83
4.1.2 Small scale positional analysis ............................................................. 88
4.1.3 Verification of 3D measurement performance ..................................... 94
4.1.4 Comparison of measurement results with a commercially available
system .............................................................................................................. 95
4.1.5 Demonstration of capability in the measurement of translucent materials
.............................................................................................................. 97
Chapter 5 Laser texturisation of glass superstrate for enhanced solar cell
performance .............................................................................................................. 98
5.1 Laser processing and parasitic light absorption losses ................................ 98
5.2 Design of Experiments and Response Surface Methodology (RSM) ......... 99
5.3 ANOVA Analysis ...................................................................................... 101
5.4 SEM and EDXS Characterisation ............................................................. 102
5.5 Effect of HF etching on absorption of light............................................... 109
5.6 Efficiency vs angle of incidence................................................................ 113
Chapter 6 Characterisation for Quantum Dot Solar Cells .................................. 115
6.1 Absorption characteristics of alkyl-capped silicon quantum dots ............. 115
6.2 Absorption characteristics of alkyl-capped silicon quantum dots on laser
textured superstrates ............................................................................................. 117
6.3 FIB serial sectioning for the 3D reconstruction of macroporous structures ....
................................................................................................................... 119
6.4 3D reconstruction and morphological analysis of the carbon monolith .... 120
6.5 3D reconstruction and morphological analysis of the silica-based monolith ..
................................................................................................................... 122
Chapter 7 Conclusions ........................................................................................ 125
7.1 Future Work .............................................................................................. 130
References ................................................................................................................ 133
Appendix A : Positional Analysis ............................................................................ i
Appendix B : RSM Procedure ............................................................................. xiv
Appendix C : Dial Gauge Technical Specifications ............................................ xvi
Appendix D : ISO 230-2 ................................................................................... xviii
Appendix E : Fused Quartz Datasheet .............................................................. xxiii
X
Appendix F Channel Dimension Analysis ........................................................ xxiv
Appendix G : 3D Analysis ................................................................................ xxvii
XI
List of Figures
Figure 1: Flowchart and structure of the thesis ............................................................ 6
Figure 2 - The Electromagnetic spectrum (source: Wikimedia Commons) ................ 8
Figure 3 - Representative circuit of a solar cell. RS and RSH represent the series and
shunt resistances respectively. ..................................................................................... 8
Figure 4 - Typical solar cell configuration. .................................................................. 9
Figure 5: Band diagram for a photovoltaic cell. (source: Wikimedia Commons) ..... 10
Figure 6 - Solar radiation spectra according to ASTM G-173-03 [14]. ..................... 12
Figure 7. Schematics of superstrates with light rays incident vertically from the top
shown in red for situations (a) with active layer (blue) on the textured side opposite
incident rays, (b) with the active layer on the untextured side; and substrates (c) with
the active layer on the textured surface, and (d) with the active layer on the flat surface.
.................................................................................................................................... 15
Figure 8. Pyramidal texture on c-Si created by etching with TMAH [18]. ............... 18
Figure 9. Perspective view of a honeycomb structure formed by chemical etching [8].
Spacing of hexagons is 14 µm. .................................................................................. 20
Figure 10. Five possible texture types (a) V-Grooves, (b) V-Grooves double sided, (c)
Pyramids, (d) Pyramids double sided, and (e) Perpendicular grooves or ‘Patch’ texture
.................................................................................................................................... 21
Figure 11: Example of a quantum dot sensitised solar cell architecture [70]. ........... 26
Figure 12. 3D non-contact optical surface profilometer. ........................................... 32
Figure 13: Main Graphical User Interface (GUI VI Front Panel) for the 3D non-contact
optical surface profilometer. ...................................................................................... 33
Figure 14. Flowchart indicating operation of the 3D non-contact optical surface
profilometer control software. .................................................................................... 33
Figure 15. Images of the (a) high resolution optical system and vertical actuator, and
(b) XY positioning stage with stepper motors. .......................................................... 36
Figure 16. Raw data processing procedure. Both a topographical height map and a
high resolution fully focused image are created. ....................................................... 38
Figure 17. 3D surface display of laser processed grooves in transparent glass. The
surface has been shaded blue to improve visibility. Measurement area is 310 µm by
232 µm. ...................................................................................................................... 38
XII
Figure 18: (a) A representation of the actual surface being measured. (b) The
components of the surface profile according to ISO 4287. ........................................ 40
Figure 19. Roughness profile is obtained using the ISO 4288 standard for profile
correction.................................................................................................................... 41
Figure 20 - PYSER-SGI PS78 Stage micrometer used for calibrating the optical plane
of the microscope system. .......................................................................................... 42
Figure 21: Mitutoyo 513-405E Dial Gauge when performing point to point
measurements on the X-axis of the 3D Optical Profilometer. ................................... 45
Figure 22: The measurement procedure was performed according to the following
sequence. .................................................................................................................... 46
Figure 23: SEM image of a 0.75 mm milling tool utilised for manufacturing the
measurement artefact. ................................................................................................ 50
Figure 24 - Wide field Backscatter Electron SEM image of machined brass sample 50
Figure 25 - Secondary Electron SEM image of end face of a pillar on the machined
brass sample. Machining debris is visible at the end face of the measurement artefact.
.................................................................................................................................... 51
Figure 26 - 3D surface measurement of the machined brass sample using a Wyko
NT1100 white light interferometer ............................................................................ 52
Figure 27 - Cross sectional Wyko NT1100 measurement data of machined brass
sample. Step height is measured as 103µm................................................................ 53
Figure 28 - 3D surface profile of the machined brass sample using a DEKTAK V200Si
Stylus profilometer ..................................................................................................... 54
Figure 29 - Cross sectional profile of machined brass samples acquired using
DEKTAK V200Si Stylus profilometer ...................................................................... 54
Figure 30: Mitutoyo Rectangular Gauge Block Set (Grade 1) .................................. 55
Figure 31: Step height of nominal 500µm Mitutoyo Rectangular Gauge Block. The
WLI measurement verified the dimension of the block to be 512µm. ...................... 56
Figure 32: a) 3D view of measurement, b) Fully focused image from a step height
measurement, c) the height map from this measurement, d) the areas of high un-
repeatability and noise due to steep flanks. ................................................................ 59
Figure 33. Graphic illustration of pulse period, duration and peak power within pulse
width. .......................................................................................................................... 61
Figure 34. Central Composite Design of Experiments .............................................. 63
Figure 35. Integrating sphere setup for absorption measurements. ........................... 66
XIII
Figure 36. Cut-away view of the integrating sphere setup......................................... 67
Figure 37: Diagram showing the SEM measurement locations on a cross sectional
view of laser micro-machined channel. ..................................................................... 69
Figure 38: Diagram showing the SEM measurement locations for the measurement of
the Width (top) parameter. Top surface is represented by red line, and measurement
locations are shown in orange. ................................................................................... 70
Figure 39. a) Angle of incidence of 90°, and b) angle of incidence of 10°. ............... 71
Figure 40. Mounting device constructed to hold textured superstrate and encapsulated
solar cell in contact. The rotational stage allowed for measurements of varying angles
of incidence to be performed...................................................................................... 71
Figure 41: The AFM image was taken on the Si-QDs cast onto mica. From the profile
along the line (a), one can find out the height of clusters was about 10 nm, which is
equivalent to the height of two layers of Si-QDs, as shown in (b) [115]. .................. 74
Figure 42: Sketch of the sample/FIB-SEM system set-up for serial sectioning and
imaging. ...................................................................................................................... 78
Figure 43 Image captured by SEM after FIB sectioning of the carbon monolith. ..... 79
Figure 44 Image captured by SEM after FIB sectioning of the silica monolith. ....... 80
Figure 45 From L to R: Outline of the pores, filling of pores and binarisation applied
to the image stack obtained for the carbon monolith. ................................................ 81
Figure 46: Large scale positional analysis along X axis. ........................................... 84
Figure 47: Large scale positional analysis along Y axis. ........................................... 85
Figure 48: Large scale positional analysis along Z axis. ........................................... 86
Figure 49: Small scale positional analysis along X axis. ........................................... 89
Figure 50: Small scale positional analysis along Y axis. ........................................... 90
Figure 51: Small scale positional analysis along Z axis. ........................................... 91
Figure 52: Measurement recorded using a Keyence VHX-5000 series 3D microscope.
.................................................................................................................................... 95
Figure 53: Fully focused true colour image of the machined brass sample ............... 96
Figure 54: 3D representation of the surface profile of the machined brass sample. .. 97
Figure 55: 3D measurement profile of laser micro-machined grooves in fused quartz
glass. Axis values are in nm. ...................................................................................... 97
Figure 56. (a) Effect of Traverse Speed with PRF=400Hz and Duty Cycle=6%. (b)
Effect of Duty Cycle with PRF=400Hz and Traverse Speed=8.333mm/s. ............... 99
XIV
Figure 57. Values predicted by the model equation against the actual values. Points
located along the line indicate the model provides an acceptable fit. ...................... 100
Figure 58. RSM surface plot showing the correlation between the significant input
parameters and the resulting parasitic absorption in the processed material
(PRF=280Hz). .......................................................................................................... 100
Figure 59. Top down view of laser micro-machined channels in Fused Quartz. (a)
Sample 21 with low absorption properties (PRF=400Hz, Speed=8.3mm/s, Duty
Cycle=2%). (b) Sample 9 with high absorption properties (PRF=400Hz,
Speed=1.7mm/s, Duty Cycle = 6%)......................................................................... 103
Figure 60. Cross sectional view of the groove texture.(a) Sample 16 with the maximum
(duty cycle 6%, PRF 160Hz, speed 5mm/s) and (b) Sample 21 with the minimum
texture depths achieved by laser ablation (duty cycle 2%, PRF 400Hz, speed
8.33mm/s). ............................................................................................................... 103
Figure 61: Average channel depths. ......................................................................... 104
Figure 62: Average channel widths at the top surface. ............................................ 105
Figure 63: Average channel width at 50% depth. .................................................... 106
Figure 64: 3D surface profile of sample 21, 70-80 micrometres in depth, as captured
by the 3D optical profilometer. ................................................................................ 107
Figure 65. EDXS results for (a) Sample 21 with low absorption, (b) Sample 9 with
high absorption, and (c) untextured Reference Sample ........................................... 108
Figure 66. The effect of hydrofluoric acid (40%) etching on the absorption of light at
the surface. ............................................................................................................... 109
Figure 67. An electron microscope image of the surface of sample 21 before acid
etching. It is possible to see a large amount of re-cast material. Visible feature sizes
range from approximately 400 nm to several microns............................................. 110
Figure 68. (a) Surface of sample 21 after 5 minutes HF etching, and (b) Surface of
sample 21 after 10 minutes etching. ......................................................................... 110
Figure 69. Top down view of the surface morphology of laser ablated channels (duty
cycle 4%, PRF 400Hz, speed 1.67mm/s). unetched (a), after etching for 10 minutes
(b) and after etching for 20 minutes (c). .................................................................. 111
Figure 70. Absorption of AM1.5G spectrum by sample 21 with respect to etching time
(data fitted with 6th order polynomial trend line for clarity). It can clearly be seen that
etching causes a shift in absorption towards the red end of the spectrum. This can be
XV
directly correlated to the elongation of nanostructures on the surface due to acid
etching. ..................................................................................................................... 112
Figure 71. EDXS spectrum of sample 21 after etching with HF. Composition is close
to that before HF etching (Figure 65(a)) .................................................................. 112
Figure 72. I-V curves for solar cell at the acute angle of 10° incidence. untextured
reference (solid), sample 20 (dotted), and sample 21 (dashed). ............................... 113
Figure 73. Solar cell efficiency with different angles of incidence for the untextured
reference (solid), sample 20 (dotted), and sample 21 (dashed). ............................... 114
Figure 74: Optical microscope image of nanocrystalline silicon on substrate shortly
after drop deposition. Bubbles indicate that the solvent is still present. .................. 116
Figure 75: Light absorption characteristics of Si QDs after 1 deposition (green), 2
depositions (blue), 3 depositions (red). The absorption of the glass and the solvent
have been subtracted from this data. ........................................................................ 117
Figure 76: Light absorption characteristics of Si QDs on laser textured superstrates
after drop casting. Untextured reference (blue), Sample 19 (red), Sample 20 (green),
and Sample 21 (violet).The absorption of the glass and the solvent have been
subtracted from this data. ......................................................................................... 118
Figure 77: Spectra for untextured reference (blue) and Sample 19 (red). The spectra
have been reference matched from 520 nm to 1000 nm. ......................................... 119
Figure 78: 3D reconstruction of the porous carbon monolith .................................. 121
Figure 79: Mercury porosimetry results showing pore size distribution in the carbon
monolith [122]. ......................................................................................................... 122
Figure 80: 3D reconstruction of the porous silica monolith (Monotrap). ................ 122
Figure 81: Mercury porosimetry results showing pore size distribution in the silica-
based monolith [122]................................................................................................ 123
XVI
List of Tables
Table 1: Outline of physical and atmospheric conditions as detailed in ASTM G-173-
03 [12]. ....................................................................................................................... 12
Table 2. Texture performance in reducing reflection from silicon solar cells ........... 19
Table 3. Optical system specification ........................................................................ 34
Table 4. Motion control system specification ............................................................ 35
Table 5: Selection of Filter parameters according to ISO 4288 ................................. 41
Table 6: Analysis of large scale movements using Mahr 810SW dial gauge ............ 44
Table 7: Analysis of small scale movements using Mitutoyo 513-405E dial gauge . 45
Table 8: Position variation from the nominal point-to-point interval. ....................... 46
Table 9: Target positions for Large Movement analysis. All positions in millimetres.
.................................................................................................................................... 47
Table 10: Target positions for Small Movement analysis. All positions in millimetres.
.................................................................................................................................... 47
Table 11: Measurement results of brass sample step height taken by the various
instruments ................................................................................................................. 53
Table 12: Design of Experiments with specified sampling intervals ......................... 58
Table 13. Laser parameters held constant during the experiments ............................ 62
Table 14. Laser process parameters used for DOE .................................................... 63
Table 15: Laser micromachining parameters for each sample produced in the DOE.
.................................................................................................................................... 64
Table 16. EDXS compositional analysis parameters ................................................. 68
Table 17: Acquisition details for carbon monolith .................................................... 78
Table 18: Acquisition details for silica monolith ....................................................... 78
Table 19: Results from large scale positional analysis along X axis ......................... 87
Table 20: Results from large scale positional analysis along Y axis ......................... 87
Table 21: Results from large scale positional analysis along Z axis ......................... 87
Table 22: Results from small scale positional analysis along X axis ........................ 92
Table 23: Results from small scale positional analysis along Y axis ........................ 92
Table 24: Results from small scale positional analysis along Z axis ......................... 92
Table 25: Accuracy results from the 6 sub-run measurements .................................. 94
XVII
Table 26. The values of the Regression Co-Efficients for the quadratic response
surface model shown in Figure 58. .......................................................................... 101
Table 27. Contribution of the most significant factors to the quadratic response surface
model. ....................................................................................................................... 101
Table 28: R-squared values for the RSM model ...................................................... 102
Table 29. Comparison of samples with highest and lowest light absorption properties.
.................................................................................................................................. 108
Table 30 Comparison of surface composition before and after HF acid etching..... 113
Table 31. Parameters and results for I-V measurement with light incident at 10° to
surface. ..................................................................................................................... 114
Table 32: Large scale positional analysis of X axis using Mahr 810SW dial gauge
indicator (1-5)............................................................................................................... ii
Table 33: Large scale positional analysis of X axis using Mahr 810SW dial gauge
indicator (6-10)........................................................................................................... iii
Table 34: Large scale positional analysis of Y axis using Mahr 810SW dial gauge
indicator (1-5).............................................................................................................. iv
Table 35: Large scale positional analysis of Y axis using Mahr 810SW dial gauge
indicator (6-10)............................................................................................................. v
Table 36: Large scale positional analysis of Z axis using Mahr 810SW dial gauge
indicator (1-5).............................................................................................................. vi
Table 37: Large scale positional analysis of Z axis using Mahr 810SW dial gauge
indicator (6-10)........................................................................................................... vii
Table 38: Small scale positional analysis of X axis using Mitutoyo MT513-405E dial
gauge indicator (1-5) ................................................................................................ viii
Table 39: Small scale positional analysis of X axis using Mitutoyo MT513-405E dial
gauge indicator (6-10) ................................................................................................. ix
Table 40: Small scale positional analysis of Y axis using Mitutoyo MT513-405E dial
gauge indicator (1-5) .................................................................................................... x
Table 41: Small scale positional analysis of Y axis using Mitutoyo MT513-405E dial
gauge indicator (6-10) ................................................................................................. xi
Table 42: Small scale positional analysis of Z axis using Mitutoyo MT513-405E dial
gauge indicator (1-5) .................................................................................................. xii
Table 43: Small scale positional analysis of Z axis using Mitutoyo MT513-405E dial
gauge indicator (6-10) .............................................................................................. xiii
XVIII
Table 44: Results from channel depth analysis ....................................................... xxiv
Table 45: Results from channel width (top) analysis ............................................... xxv
Table 46: Results from channel width (50%) analysis ........................................... xxvi
XIX
Nomenclature
Term Description
Duty Cycle (%)
The ratio of the pulse duration to the pulse period. A duty
cycle of 100% would indicate that the output is
continuous.
Pulse Duration (s) The amount of time the laser beam is actually irradiating
energy.
Pulse Period (s) The amount of time from the start of one pulse to the start
of the next pulse.
Pulse Repetition
Frequency (PRF) (Hz)
The frequency of the laser pulses. The inverse of pulse
period.
AM1.5G Spectrum
American Society for Testing and Materials (ASTM)
Terrestrial Reference Spectra for Photovoltaic
Performance Evaluation
Spectral Irradianceλ %
The percentage of the total incident radiant energy per
wavelength of the incident spectrum. e.g 420 nm
represents 0.16% of the total radiant energy between 42
0nm and 1000 nm.
Si-QD Silicon Quantum Dots, also known as silicon nanocrystals
1
Abstract
Laser texturisation of photovoltaic module superstrates for enhanced light
trapping performance
David John Moore
In order to increase the efficiency of solar cell modules it is necessary to make the
optimum use of light incident upon them. Much research has been conducted to
improve light absorption through front surface texturing and light trapping schemes.
Laser light is commonly used in industry for various applications including marking
and texturing. By controlling laser parameters, it is possible to tailor macro and micro
structures in most materials. A CO2 laser operating at 10.6μm wavelength was used to
produce grooved textures in fused quartz material with a view to its usage as a cover
glass on top of the photovoltaic cell surface. With correct texturing it is postulated that
increased energy absorption can be promoted due to trapping of light within the
photovoltaic cell due to total internal reflection and enhanced optical path lengths.
Analysis of the effects of the laser parameters on the texture geometry and surface
morphology was performed through a combination of cross-sectioning and scanning
electron microscopy. Transmission spectra through the textured glass samples were
recorded, and transmission through the glass was improved for most samples after acid
etching. It was found that for acute angles of incidence of wavelengths of natural
sunlight upon the cells, greater coupling efficiencies were achieved compared to flat
surfaces, due to the increased light trapping effect. The main contributions of this work
include examination and quantification that indicate the laser textured solar
superstrates can increase the light trapping effect within silicon solar cells and that an
enhanced light trapping can be achieved when silicon quantum dots deposited directly
on a textured superstrate. Another two important contributions are found in the
development of characterisation methods and analyses of microstructures relevant to
light trapping in current and emerging solar cell technologies. These include the
design, fabrication, development, and verification of a newly designed and
commissioned depth-from-focus based optical profilometer with which new results in
the metrology of translucent surfaces with micro-scale roughness are presented.
Software to analyse serial FIB-SEM sections of monolithic porous microstructures
was also developed. The results gained from these characterisation methods have
allowed, and it is postulated will in the future provide, a more detailed understanding
of the light trapping process with various microstructures.
2
Chapter 1 Introduction
Solar cells operate on the principle of the photoelectric effect; the conversion of light
energy to electrical energy. While there are many variations of solar cells, this energy
conversion can be achieved by illuminating a semi-conducting material with light to
generate an electron-hole pair, also called an exciton [1]. There are various materials
that are used as light absorbers in order to achieve the photoelectric effect, the most
commonly used today being silicon, mainly due to its widespread availability, the
availability of existing mature processing and handling technology from the
electronics industry, and also its band-gap suitability for absorption of the solar
spectrum.
While the efficiency of solar cells has greatly increased in recent years, the cost per
Watt peak is still too high for photovoltaics to be considered more widely as a cost
effective means of energy production. In order to reduce of the cost per Watt peak of
the solar cells, it is necessary to achieve one of the following - higher solar cell
efficiencies or lower manufacturing costs. While current high performance silicon
solar cells obtain efficiencies upwards of 20 percent, there are still significant
opportunities for further improvements [2]. The Shockley-Queisser limit states that
the theoretical maximum efficiency of a single junction silicon solar cell is 33% under
AM1.5 Global radiation. In order to achieve this goal, the solar cells must make
optimum use of the light incident upon them. While there has been much research into
the improvement of semiconductor fabrication techniques, there still remains an
opportunity to better improve the optical performance of the solar cells. The optical
performance of silicon solar cells is limited in two ways - silicon is a relatively poor
absorber of near-infrared light, and silicon also has a high refractive index meaning
there are large losses due to Fresnel reflection at the front surface [3]. Improved light
absorption characteristics have been observed in silicon solar cells through the use of
anti-reflective coatings, among which silicon nitride (Si3N4) is among the most
popular [4, 5]. However, these anti-reflective coatings operate optimally only in
certain ranges of wavelengths of light and at certain angles of incidence of solar
radiation. Other steps implemented to improve the optical performance include direct
texturisation of the front silicon surface in the solar cell. This can be achieved through
3
various methods, the most common of which is chemical etching and is most effective
on single crystal silicon (c-Si). While single crystal silicon can be used to make the
cells with the highest efficiencies, due to the lack of internal lattice defects, it is also
the most expensive form of silicon to produce due to the energy intensive nature of
the manufacturing process [6]. Also, more recently thin-film solar cells and quantum
dot based solar cells are coming to the forefront of research and markets, and the direct
texturisation of these layers is difficult using existing manufacturing methods. While
thin-film solar cells in particular offer an excellent opportunity to reduce the
manufacturing costs of the cells, the associated optical losses limit the maximum
performance obtainable with this architecture. In order to reduce the optical losses and
to simultaneously reduce the manufacturing costs associated with the solar cell, it is
necessary to develop a method of improving light absorption characteristics which is
suitable for use with wafer based bulk solar cells and thin-film type solar cells. It is
postulated that an optimised surface texture can increase the ability of the solar cell to
absorb incident light more effectively, in a process known as light trapping. Quantum
dot based solar cells show great promise in being able to overcome the Shockley-
Queisser limit, and the band-gap of silicon quantum dots can be controlled by
controlling the size of the dot [7]. While the architecture of quantum dot based solar
cells is somewhat different to the traditional approach in wafer based and thin-film
silicon solar cells, there are emerging architectures that utilise micro- and nano-
structured materials for improved light trapping performance.
Objectives of this work
There are two main objectives for this research. Firstly, the aim is to investigate
fundamentally the use of lasers to texturise on the micro scale the front cover glass
superstrate as would be used in a conventional solar module. From previous
simulations performed in literature, it is estimated that the amount of light trapped in
a solar cell by trapping secondary and subsequent surface reflections can increase the
optical thickness of the solar cell by up to 40 times [8]. This is especially important
for thin-film architectures when considering that light with photon energies close to
the band-gap of the semiconductor are weakly absorbed in thin layers. The use of
lasers to create the textures offers excellent flexibility in terms of process parameters
to achieve a range of surface textures for use with the solar cells. Lasers also offer a
4
non-contact process which is highly scalable in terms of production speed and
throughput, thus offering the possibility to reduce the manufacturing costs and
simultaneously increase the efficiencies of the solar cells.
In the real world use of solar cells, modules are often installed in fixed orientations,
such as on the roofs of houses or facades of buildings. The quoted efficiencies of solar
cells are those under ideal testing conditions of AM1.5G in a laboratory environment.
However, in day-to-day operations the solar module will experience different
conditions and different angles of incidence of sunlight. Thus the motivation behind
this research is to also encompass an investigation into how the angles of incidence of
light will affect the solar cell performance, and how different superstrate textures could
be used to improve light coupling into the solar cell. As the angle of incidence of light
decreases from normal to parallel, it is postulated that the texture on the surface will
assist the coupling of more light to the surface of the cell within the solar module, as
opposed to having a flat superstrate surface where a large amount of light is lost due
to reflection.
The objective of carrying out this research using glass as the front cover superstrate
lies in the properties of the material itself, and also in its suitability for the application.
Glass is currently the preferred choice for silicon based solar module manufacturers
due to its low cost and wide availability, and its beneficial mechanical and optical
transmission properties.
The second objective of this work is to investigate the characterisation methods of
microstructures used in solar cells. While there are several established methods for the
characterisation of surface nanostructures, the measurement of microstructures (on the
order of 1 micrometre to 200 micrometres) still presents some challenges.
Microstructures with high aspect ratios in this size range are outside the measurement
capability of stylus instruments such as Atomic Force Microscopes (AFM), while
significant surface roughness can also present measurement challenges to optical
instruments such as White Light Interferometers (WLI). Some types of solar cells,
particularly those based on quantum dot architectures, depend on semi-conducting
materials for operations. This work also investigates the use of an advanced Electron
Microscopy technique called Focussed-Ion Beam Serial Sectioning and Imaging, in
5
order to assess the 3D structure of macroporous monolithic materials. It is believed
that this work will contribute better to the understanding of the suitability of these
techniques for solar cell material and microstructure analysis. An additional benefit of
this work lies in the fact that these characterisation methods allow surfaces of the
structures topologies to be exported by the user for use in optical simulations such as
ray-tracing, which would greatly add to the realism of simulations within the field of
photovoltaics.
6
Chapter 1
Introduction
Chapter 2
Literature Review
Chapter 3
Materials & Methods
Chapter 4
Optical Profilometer
Results
Chapter 5
Laser textured glass on
Solar Cells Results
Chapter 6
Analysis of Silicon
Quantum Dot Solar
Cell components
Chapter 7
Conclusions
References
Appendices
Study objectives
Thesis outline
Solar Cell operation
Light absorption in Solar Cells
Surface texturisation of Silicon Solar Cells
Design and Verification of 3D Profilometer
Laser texturisation of glass superstrates
Measurement of light absorption losses in textured glass
Solar cell performance at varying angles of incidence
Silicon Quantum Dot deposition
Focused-Ion Beam Serial Sectioning
Axis performance analysis
Step height measurement verification
Demonstration of capability
Laser texturisation
Measurement and ANOVA of absorption losses
HF etching to reduce absorption losses
Solar Cell performance with textured superstrates at
varying angles of incidence.
Absorption spectrum of alkyl capped Silicon Quantum dots
Focused-Ion Beam Serial Sectioning of micro and nano
porous media
Verification of Optical Profilometer Performance
Performance of Solar Cells with texutred superstrates
Focused-Ion Beam Serial Sectioing and Quantum dot solar
cells
Further work and recommendations
Figure 1: Flowchart and structure of the thesis
7
Chapter 2 Literature Review
2.1 Introduction to solar cells
What is a solar cell?
A solar cell, also called a photovoltaic cell, is a solid state device used for the
conversion of light energy to electrical energy. The device makes use of the physical
and chemical phenomenon known as the photovoltaic effect. Photovoltaic cells utilise
the properties of semiconductors in certain configurations to effectively absorb
photons of light incident upon the material and convert the energy of the incident
photons into electrical current [9]. While there are various possible solar cell
configurations, the most commonly used solar cells type is based on the use of silicon
[5]. Silicon is a well understood widely used semiconductor material in the electronics
industry, is widely available, and due to the accessibility to existing fabrication
techniques lends itself well to creating a cost effective solution in the production of
photovoltaic devices [6]. Importantly, silicon also absorbs electromagnetic radiation
from the near-infrared end of the spectrum to the UV, which represents coverage of a
significant band of electromagnetic solar energy incident upon Earth [1]. It is
estimated that silicon based solar cells account for approximately 92% of the global
solar cell market in 2014 [2]. Great improvements in silicon solar cell efficiency have
been made in the last decade, through improved fabrication techniques, and better
management of factors that can negatively influence solar cell performance.
2.2 The mechanics of the generation of photocurrent
The electromagnetic spectrum describes a range of wavelengths (or frequencies) of
electromagnetic radiation, as displayed in Figure 2. From a solar cell perspective, one
is typically concerned with electromagnetic radiation ranging from the near –infrared
(NIR) to the ultraviolet (UV) [4, 5]. There are two main reasons for this. Firstly, the
bulk of the solar radiation reaching sea level on Earth consists of wavelengths within
this region of the electromagnetic spectrum. Secondly, the reason for this is due to the
quantum mechanics of the photovoltaic effect. The photon energies of photons within
the NIR to UV spectrum are well matched to the band-gap energies of common
semiconductors, meaning that it becomes relatively efficient to generate photocurrent
within this range. Photons of a particular wavelength also have a corresponding photon
8
energy which inversely proportional to the wavelength. For example, a photon with a
wavelength of 500 nm has an equivalent photon energy of 2.48eV, while a photon
with a wavelength of 700 nm has an equivalent photon energy of 1.77eV [10].
Figure 2 - The Electromagnetic spectrum (source: Wikimedia Commons)
The photovoltaic effect operates on the principle of utilising the properties of a P-N
junction to achieve charge separation. From an electrical point of view, a solar cell can
be represented as an ideal current source circuit with a series and shunt resistance, and
behaviour identical to that of a traditional diode [1]. An example of the equivalent
circuit can be seen in Figure 3.
Figure 3 - Representative circuit of a solar cell. RS and RSH represent the
series and shunt resistances respectively.
9
Traditional diodes are also comprised of a P-N junction [11]. This type of junction is
formed by bringing into contact two differently doped semiconductors. Typically, in
the case of silicon, the p-type semiconductor is doped with Boron, and the n-type
semiconductor is doped with Phosphorous. There are several effects of doping in this
manner. Firstly, the carrier concentrations of the intrinsic (pure) semiconductor are
modified, meaning that the p-type semiconductor has excess holes, and the n-type has
excess electrons. The amount of doping therefore modifies the conductivity of the
semiconductor, which can be beneficial in reducing the series resistance of the solar
cell to maximise cell efficiency. Secondly, the P-N junction formed has the property
of only allowing current to flow in one direction. Finally, the P-N junction forms an
internal electric field and a depletion region near the boundary of the junction,
meaning is it possible to achieve change separation, and hence to generate
photocurrent by utilising the photovoltaic effect.
Figure 4 - Typical solar cell configuration.
In order to generate photocurrent in a solar cell, the incident photons of light must
energise electrons within the absorber material. When a photon is absorbed by an
atom, the energy of the photon is transferred to an electron. Typically, electrons are in
a ground (valence) state, and then become excited to a higher state on the absorption
of a photon as shown in Figure 5. However, there is a minimum amount of photon
energy required to excite the electron sufficiently so that it can contribute to
photocurrent. This minimum energy is known as the band-gap (eV), and is unique to
10
each semiconducting material [1, 3]. The band-gap has several important implications
for the performance of a solar cell.
Photons at energies below the semiconductor band-gap are not absorbed do not
contribute to photocurrent.
Photons exactly at the band-gap energy can generate photocurrent with high a
spectral response, as the photons contain just enough energy to excite the
electrons into the conduction band.
Photons with energies higher than the semiconductor band-gap are absorbed
and can contribute to photocurrent generation. However, the residual energy
(photon energy minus band-gap energy) is lost in the form of thermalisation.
This effect reduces the spectral response of the solar cell at shorter
wavelengths, limiting the possible device efficiency.
Figure 5: Band diagram for a photovoltaic cell. (source: Wikimedia
Commons)
11
2.3 Silicon as a photovoltaic material
For silicon, the band-gap energy is approximately 1.1eV, which corresponds to an
electromagnetic radiation wavelength of 1100nm. This means that silicon is very well
suited as a photovoltaic material, as its band-gap energy is well matched with the solar
radiation flux incident on the Earth [1]. As described in section 2.4, the majority of the
solar radiation flux reaching Earth is at wavelengths shorter than 1100nm, and hence
the band-gap of silicon offers a well suited balance between being low enough to allow
for absorption for the majority of solar radiation and simultaneously offering
reasonable quantum efficiency (QE) performance over the usable portion of the
spectrum.
Crystalline silicon is an indirect band-gap material, which means that in order for an
electron to move from the valence band into the conduction band a chance in electron
momentum (k-vector) must also occur in conjunction with transfer of energy from an
incoming photon above the band-gap energy [3]. This change is momentum is
normally provided by interaction with a phonon, or lattice vibration. What this means
in physical terms is that the probability of an electron transitioning from the valence
band to the conduction band is greatly reduced compared to the probability of the same
transition in a direct band-gap semiconductor. The absorption co-efficient of
crystalline silicon is dependent on the wavelength of light passing through it, as given
be the Beer-Lambert Law [3]. Despite this, silicon is by far the most widely used
semiconductor in photovoltaics. Combining the favourable band-gap characteristics
of silicon with its wide availability and the availability of mature silicon processing
technologies from the micro-electronics industry, it is clear why silicon solar cells
have been so successful on an industrial production scale.
2.4 The Air Mass (AM) standard
The Air Mass standard provides information on the Spectral Irradiance of solar
radiation; AM0 represents the extra-terrestrial spectrum, while AM1.5 Direct
represents the irradiance of solar energy on the Earth’s surface. The AM1.5 Global
(AM1.5G) spectrum utilised in this work represents the combined direct and diffuse
solar radiation reaching the Earth’s surface at specific conditions. The AM1.5G
standard is represented by ASTM G-173-03 and ISO 9845 [12, 13]. The Air Mass
spectra are shown in Figure 6.
12
Figure 6 - Solar radiation spectra according to ASTM G-173-03 [14].
There are certain specifications involved in the calculation of the AM1.5G standard.
Certain atmospheric and physical conditions are assumed. The main considerations
are outlined in Table 1.
Table 1: Outline of physical and atmospheric conditions as detailed in
ASTM G-173-03 [12].
Factor Details
Receiving
surface
Inclined plane at 37° tilt toward the equator, the surface normal
points to the sun, at an elevation of 41.81° above the horizon
Atmosphere The 1976 U.S. Standard Atmosphere with temperature, pressure,
aerosol density (rural aerosol loading), air density, molecular
species density specified in 33 layers
Atmosphere An absolute air mass of 1.5 (solar zenith angle 48.19°s)
Atmosphere Angstrom turbidity (base e) at 500 nm of 0.084
Atmosphere Total column water vapour equivalent of 1.42 cm
Atmosphere Total column ozone equivalent of 0.34 cm
Atmosphere Surface spectral albedo (reflectivity) of Light Soil as documented
in the Jet Propulsion Laboratory
13
2.5 Types of wafer based silicon solar cells
There are several manufacturing methods available for the production of Silicon solar
cells. The most common method for producing high efficiency solar cells is through
the use of single crystal silicon. Single crystal silicon (c-Si) is most commonly grown
using the Czochralski process, and produces very high quality silicon [5, 6]. However
this process is time and energy intensive, and hence this production method also
attracts higher production costs. These solar cells are used in applications where
energy density is critical such as in space applications, although improved
manufacturing methods and economies of scale have also made this type of solar cell
popular for terrestrial applications.
While single crystal based solar cells have the highest efficiencies in the Silicon PV
market, multi-crystalline solar cells make up the majority of that market, with an
estimated 56% of silicon solar cells produced in 2014 being based on multi-crystalline
silicon technology [2]. The mutli-crystalline form of silicon is significantly cheaper to
produce, however, these cells have a drawback in that their maximum efficiency is
significantly below that of cells manufactured from single crystal silicon. This is
mainly due to the fact that the grain boundaries in the multi-crystalline material act as
defects and are centres for electron-hole re-combination, essentially reducing the shunt
resistance of the solar cell and the cell efficiency [3]. There are also several methods
for manufacturing multi-crystalline solar cells. The most common manufacturing
method is the casting of melted silicon, where the grain size can be somewhat
controlled by controlling the cooling rate of the molten material. Another method,
dubbed the Ribbon method, involves the continuous casting and cooling of a thin
ribbon of silicon. This method has the potential to be the most cost effective way of
making multi-crystalline solar cells, however the drawback of this approach is the
difficulty in maintaining wafers without cracks or significant crystal defects [1].
2.6 Factors limiting the absorption of light in silicon solar cells
There are several significant factors that limit light absorption in traditional silicon
solar cells. Firstly, silicon has a high refractive index meaning that at a silicon-air
interface there is significant light loss due to Fresnel reflection [1]. In order to
overcome this limitation certain anti-reflective coatings have been developed.
Secondly, silicon is a relatively weak absorber of light in the near-infrared region of
14
the electromagnetic spectrum, near to the materials band-edge. The reason for this is
that silicon is an indirect band-gap semiconductor, meaning that along with the
absorption of a photon, an interaction with a lattice vibration (phonon) is required for
an electron to transition between the valence band and the conduction band [3]. This
means that crystalline silicon has a relatively low absorption probability for photon
energies close to the band-gap. For crystalline silicon, this means the production of
thicker solar cells, to enhance the absorption probability of lower energy photons. In
thin film solar cell designs where material thickness is limited, amorphous silicon is
often utilised as it behaves as a direct band-gap semiconductor which improves
efficacy in thin layers, however silicon in this form also has a higher band-gap (closer
to 1.75eV), which in turn has a knock on effect on the materials ability to absorb the
solar spectrum [3].
2.7 Possible device configurations for effective light trapping
There are many possible device architectures and surface texturisation techniques that
could be considered for enhanced photo-absorption. Regarding wafer based solar cells,
most research work has been focused on the direct texturisation of the actual silicon
wafer used in the solar cell. A relatively small amount of research work has
investigated the effects of solar cell superstrate or substrate texturisation, which are
utilised in front and behind the solar cell respectively. Figure 7 shows possible
configurations of solar cell modules, including locations for the active medium
positioning on textured superstrates and substrates. In thin-film and quantum dot solar
cells, the active layer is usually less than 30μm thick, meaning that this type of solar
cell must be manufactured onto a supporting superstrate or substrate [1]. Therefore, in
order to improve light trapping through texturisation it is necessary to texturise the
supporting medium through etching, laser ablation, mechanical scribing or embossing.
Possible configurations are shown in Figure 7.
15
(c)
(d)
Texturedsubstrates
(a)
(b)
Texturedsuperstrates
Figure 7. Schematics of superstrates with light rays incident vertically
from the top shown in red for situations (a) with active layer (blue) on
the textured side opposite incident rays, (b) with the active layer on the
untextured side; and substrates (c) with the active layer on the textured
surface, and (d) with the active layer on the flat surface.
16
2.8 Important factors in surface texturisation
As some light trapping approaches utilise direct texturisation of the active region of
the solar cell, there are factors that must be taken into account when designing such a
process. Not only it is necessary to ensure that the texture itself is effective in reducing
the front surface reflection, but it is also important to consider the condition of the
surface of the solar cell. One study looked into the relationship between front surface
reflectance and minority carrier lifetime within a dry etching texturised solar cell [15].
It was concluded that there is an effective trade-off between surface damage and
reflectivity. It is a common to perform a second wet-etching step after processes such
as Reactive Ion Etching (RIE), laser machining, and mechanical grooving. Dobrzanski
et al. investigated the effects of post-laser etching chemical treatment of multi-
crystalline silicon, and it was found that there is a strong relationship between high
surface damage and low cell efficiencies [16]. When surface damage was present,
short-circuit current ISC and open-circuit voltage VOC were significantly reduced. The
advantage of texturing the front surface was only observed after 20 minutes of wet
etching in a 20% potassium hydroxide (KOH) solution. Li et al. investigated the effect
of surface roughness of zinc oxide (ZnO) covered silver (Ag) layers on Corning glass
on cell efficiency [17]. In their work, nano-crystalline Si was deposited on these
substrates by hot-wire CVD, and after testing it was concluded that when a high
substrate roughness is present, the small cavities on the substrate influence negatively
the quality of the deposited silicon causing voids, thus leading to a lower VOC and
lower cell efficiencies.
2.9 Common front surface textures
2.9.1 Pyramidal texture
There are several types of textures that have been tested on solar cells for the purpose
of increasing light trapping capabilities. One of the most common types is a pyramidal
texture of the front surface, and can be achieved through a number of methods. This
structure is more regularly used in the texturisation of mono-crystalline Si as it is
desirable to have a crystal orientation of <100> to achieve a uniform pyramidal texture
[18-20]. Frequently wet or chemical etching is used to create these pyramidal
structures. This is possible by using chemical etchants such as KOH or tetramethyl
ammonium hydroxide (TMAH). The etching is anisotropic along the <100> and
17
<111> crystallographic planes thus forming a pyramidal surface topography. In one
study, TMAH and KOH were used to create a pyramidal texture [18]. It was found
that while the TMAH etch was effective, the KOH etch still produced better anti-
reflection characteristics. It was clear that the TMAH etched texture was non-uniform
and a cause of less effective results. However in another study, TMAH and KOH were
also used to etch the front surface over a period of 30 minutes at 80°C with
concentrations of 2% and 1% respectively (with 8% isopropyl alcohol (IPA)) [19].
Using TMAH led to the creation of an optimum uniform surface network of pyramids,
which when coated with a Si:N anti-reflective (AR) coating, reduced the weighted
reflection to 2.7%. There are other methods that can be used in conjunction with mono-
crystalline Si also. Reactive Ion etching (RIE) was used in one study to create hole
and columnar structures on mono-crystalline Si by masking the surface in the desired
pattern prior to etching [21]. This resulted in average reflectance values of 0.4% and
6.8% respectively. The main advantage of RIE lies in its ability to process with greater
speed, and with the use of a mask, can also be used to texturise multi-crystalline Si.
Furthermore, one study developed a novel combination of methods for the production
of additionally textured pyramidal structures though the use of plasma texturing from
a RIE source on a previously wet etched surface [22]. It was found that surface
reflection was further reduced, particularly at the blue end of the spectrum. Xu et al.
used plasma etching to form uniform conical surface structures which resulted in black
silicon [23]. These conical surface structures were shown to be particularly effective
at varying angles of incidence, considerably reducing surface reflections compared to
polished silicon. Nevertheless, other studies have simulated that the optimum structure
is a pyramidal structure with a period shorter than that of the wavelength of light
incident on it due to the benefit of the elimination of diffraction order light reflection
from such a sub-wavelength structured surface. Chen et al. used coupled-wave
analyses to simulate sub-wavelength structures, and it is reported that a pyramidal
texture on the Si surface with a period of 350nm almost completely eliminated
reflection without the aid of an AR coating, as the sub-wavelength features eliminate
almost all diffraction orders [24]. In their work, these structures were fabricated using
a combination of optical lithography and RIE, and after testing were found to have a
reflection value of approximately 2%.
18
Figure 8. Pyramidal texture on c-Si created by etching with TMAH [18].
2.9.2 Random texture
Random texture structures are commonly created on multi- or poly-crystalline Si to
reduce front surface reflection. The reason for the random appearance in comparison
with that of mono-crystalline Si is that the preferred crystalline orientation is not in
the <100> configuration, but is randomised, and thus it is not possible to wet etch in a
uniform manner to form pyramids as is possible with mono-crystalline Si. In one study
a method was developed to quickly texturise multi-crystalline Si with a randomised
layout [25]. This method used negative potential dissolution in conjunction with KOH
to etch small columns on the surface of the Si, and proved an effective way to reduce
the surface reflection significantly with a process of only two minute duration.
Schmiga et al. reported the development of 17.6% efficient tri-crystalline solar cells
which were optimised by creating crater like structures in the front surface through
sulphuric, nitric, and hydro-fluoric acid etching [26]. This resulted in a AM1.5G
weighted front surface reflection of 8.8% with an AR coating. In another approach,
Srivastava et al. utilised an Ag layer deposited onto the Silicon surface by electroless
deposition to assist an etching process using HF and H2O2 [27]. This proved a
beneficial approach, as the etching time for this process is reported to be 30 seconds,
and when comparing this with a 4 minute etch time reported by Schmiga et al., a
significant reducing in process time can be realised [26]. Interestingly, Parida et al.
utilised a very similar approach as Srivastave et al., yet found that surface reflection
19
only reduced in the 300nm to 500nm region of the solar spectrum for an etching time
of 30 seconds [28].
Table 2. Texture performance in reducing reflection from silicon solar
cells
Texture Type Method Reflectivity %
[Weighted AM1.5G]
Reference
Polished Si N/A 34 [16]
Pyramid Chemical Etch <15 [18]
Pyramid Chemical Etch <13 [19]
Pyramid Chemical Etch <15 [20]
Pyramid RIE <6.8 [21]
Pyramid RIE <8 [22]
Pyramid RIE 2 [24]
Random Chemical Etch <30 [25]
Random Chemical Etch 8.8 [26]
Random Chemical Etch <3 [27]
Random Chemical Etch <30 (estimate) [28]
Honeycomb Chemical Etch <20 [29]
Honeycomb Chemical Etch <5 [30]
Honeycomb Laser Texturisation
and Chemcial Etch
<15 [31]
Honeycomb Chemical Etch <20 [8]
Groove Laser Texturisation
and Chemcial Etch
8.3 [16]
2.9.3 Honeycomb texture
Hexagonal shaped pits etched into the surface of the Si have proven to be an effective
method for reducing front surface reflection, and is comparable in terms of optical
performance to an optimised pyramidal structure [8]. The honeycomb shape can be
produced by masking the surface and etching using an appropriate technique such as
Reactive Ion etching or wet chemical etching [21, 29, 30]. It is detailed that with this
kind of texture, an efficiency of 19.8% was obtained for a multi-crystalline solar cell.
20
In a further investigation, Abbott and Cotter used a Q-switched neodymium-doped
yttrium aluminium garnet (Nd:YAG) pulsed laser, with a wavelength of 1064 nm, to
form pit shaped structures in multi-crystalline Si, similar in morphology to a
honeycomb structure [31]. These pits were deep and at the front surface were about
20 μm in diameter. It was necessary to remove surface damage caused by the laser
irradiation and ablation using a two stage wet etch; this was conducted with a 25%
sodium hydroxide (NaOH) solution for 3 minutes at 40 °C followed by an isotropic
etch with a 30:10:4 mixture of nitric, acetic, and hydrofluoric (HF) acids for a further
3 minutes. Several different combinations of texture were tested, including single
sided and double sided laser textures. For incident light at wavelengths between 300
nm and 1000 nm, almost no difference in reflectance characteristics was observed,
however, from 1000 nm to 1200 nm the double sided texture showed superior
performance.
Figure 9. Perspective view of a honeycomb structure formed by chemical
etching [8]. Spacing of hexagons is 14 µm.
2.9.4 Grooving
The term grooving is used to describe the technique of producing straight channels on
the Si surface. This presents another possibility for the reducing surface reflection.
Relative simple mechanical and laser methods can be used instead of etching
21
techniques to produces these structures. In one study, a 50 W fundamental frequency
Nd:YAG laser was used to produce two different grooved surface structures on
different sample surfaces; one was with parallel grooves and the other with a cross-
hatched square grid pattern [16]. After a 20 minute post-process wet etch in KOH to
remove laser induced surface damage, a reduction in reflection was observed from
34% for an untextured surface to just 8.3% for the cross hatched pattern, while cell
efficiency was increased by 1.9%. This study demonstrated the effectiveness of using
lasers in the texturisation of multi-crystalline Si.
(a) (d)
(b) (e)
(c)
Figure 10. Five possible texture types (a) V-Grooves, (b) V-Grooves double sided,
(c) Pyramids, (d) Pyramids double sided, and (e) Perpendicular grooves or ‘Patch’
texture
22
2.10 Laser micromachining in solar cell design
Lasers are frequently used tools in solar cell manufacturing [32, 33]. Edge isolation,
scribing, selective ablation of layers (through the use of particular wavelength laser
for specific materials), marking, and drilling. The most commonly used industrial
lasers in solar cell manufacturing are Nd:YAG and Nd:YVO4 lasers, as they are
reliable and operate at high power which can increase process throughput [34]. Several
studies have previously looked into the direct writing technique for the ablation and
texturisation of Si solar cells with these types of lasers [16, 31]. Furthermore, other
research has been carried out on the ablation of Si using short pulse lasers such as UV
nanosecond and Ti:Sapphire femtosecond lasers [35, 36]. These investigations
developed processes for efficiently structuring thin-film solar cells for the purpose of
developing more efficient module solar modules and production methods of same.
2.10.1 Design of solar cells on texturised substrates
Rim et al. performed an interesting investigation into the fabrication of solar cells on
V-shaped grooved substrates [37]. In their work, the thickness of the active layer of
the solar cell was many times less than the profile of the texture. This meant that the
active layer morphology was similar to that of the textured glass. The light trapping
characteristics of this design were analysed for different types of active layers
including amorphous and micro-crystalline Si, which were only 300nm and 1.6μm
thick respectively. It was noted in their work that the V-shaped grooved surface was
very effective on two counts; firstly the light trapping characteristics of the cell were
improved and secondly, the texture significantly increased the surface area of the
active layer on the substrate. Kang et al. also carried out an investigation in to the
optimum textured superstrate design for inclusion in building integrated photovoltaics
[38]. The study takes into account the average angle of the sun in order to optimise the
light trapping effect. Asymmetric micro-textures were manufactured though the
production of a metallic mould, from which PDMS casts were produced. These casts
were subsequently attached to the cover glass of photovoltaic panels mounted
vertically on a building façade. The study reported a significant increase in average
cell efficiency under these conditions, with an average annual energy production
enhancement of between 12.9% and 15% respectively.
23
2.10.2 Micro-structuring and Nano-structuring of glass substrate with laser
Laser ablation can be used for the production of micro and nano-scale structures on
the surface of glass. For the ablation of glasses and ceramics in particular, CO2 lasers
have proven to be very effective as glasses absorb the corresponding wavelength of
10.6μm very strongly (i.e. they have a high quantum efficiency) [34, 39]. Much
research has been carried out in the field of laser machining of polymers and glasses
particularly for the purpose of micro-fluidics. Klank et al. studied microstructures
produced by CO2 laser micromachining of PMMA for the production of micro-fluidic
channels [40, 41]. Another study completed an experimental investigation into the
fabrication of micro-fluidic channels in a quartz glass block using a Nd:YAG laser
[42]. In many applications it is desirable to use shorter pulse and shorter wavelength
lasers as they offer tighter focussing and smaller heat affected zones (HAZ), meaning
there is a higher degree of control over ablation [39]. Some studies have directly
investigated the effect of pulse duration on the quality of the geometries fabricated via
micromachining [35, 43]. It is indicted from these studies that it is possible to create
smaller features with shorter pulse widths and low average powers. However, there is
a trade-off between pulse width and quality. Markillie et al. discovered that pulse
widths of 5μs created deeper craters in silica glass than 100μs pulse widths for the
same fluence (J/cm2) settings. This was due to the higher peak power for the shorter
laser pulse. However, it was noted that the shorter pulse also led to large amounts of
debris from melt ejection. By selecting the correct type of laser with suitable pulse
widths it is possible to have great control over the resulting micro and nano-structures.
Issa et al. investigated the use of a 1.5kW CO2 laser to perform micromachining in
soda-lime glass [44]. It was shown experimentally that by controlling the parameters
of the laser, it was possible to create an atlas for resulting channel widths, depths, and
shape from a single pass with prescribed laser settings. Nikumb et al. also carried out
an investigation into precision laser micromachining in glass using a frequency tripled
Nd:YAG, a fundamental Ti:Sapphire, and a fundamental Nd: YVO4 [45]. It was found
that significantly better quality micromachining could be performed by using pulse of
short duration. Both Kazansky et al. and Crawford et al. reported the creation of sub-
micron structures, in bulk silica and Si respectively, using an infra-red femtosecond
laser beam [46, 47]. These structures were detailed as having periodic structures as
small as 140nm.
24
2.11 Computer simulation for optimum texture type
Much work has been carried out in the area of computer simulation of light trapping
structures. Chung et al. carried out an investigation to find the optimum properties for
light trapping using pyramidal structures [48]. For this analysis a simulation was
performed to predict the absorption of photons in Si solar cells with respect to the
incident angle and pyramidal texture. It was found that a high incident angle resulted
in lower cell efficiencies. In a study by Campbell and Green, the light trapping
properties of different surface structures are presented [49]. This investigation
suggests that the optimum pyramidal structure for anti-reflection and light trapping
characteristics should contain a slight randomisation. It was also proposed that the
most efficient structure was a double sided structure (i.e. both sides of the support
structure being textured) with a cross-hatched pattern. From simulation results of
Campbell and Green, it was noted that the light trapping probability of such a double
sided textured design is optimal for Si wafers above 100μm thickness [50]. However,
these simulations only concern Si wafer solar cells which are many times thicker than
thin film devices (which are typically 5 µm or less). These thin films conform more
closely to the surface profile of the support structure. Nevertheless, the same theory
can be applied to thin film devices, with the exception that the support is texturised
before deposition of the active Si layers, as performed by Li et al. [17]. Sanches-
Illescas investigated the effect of pyramidal and cosine shaped structures on solar
panel cover glass on the light trapping performance of the panel, and found that the
textures can provide favourable transmission characteristics, particularly at oblique
angles of incidence [51].
2.12 Thin-film solar cell designs
The function of glass in solar energy systems was presented in detail in 2008 [52]. For
the fabrication of thin-film solar cells it is necessary to use a substrate or superstrate
to provide mechanical support for the thin active layers. In order to reduce
manufacturing costs, it is desirable to use a superstrate which will form an integral
part of the finished solar cell. Therefore, glass makes the ideal candidate superstrate
for thin-film solar cell production. Deubener et al. discuss various concepts in thin-
film solar cell design, such as texturing the front surface of the glass and texturing the
back side [52]. It is considered that texturisation of the front side would lead to dirt
25
accumulation under standard outdoor solar cell deployment conditions. Although it
may be possible that, with careful texturisation, a hydrophobic surface could be
produced to work in conjunction with the light trapping capabilities afforded by the
texturisation [53]. Texturisation of the back and front sides of the glass however, could
provide a promising solution as this would allow for a cell architecture and structure
similar to those optimised though simulation by Campbell and Green [49]. An
effective architecture, crystalline silicon on glass (CSG), is discussed in a study
presented by Green et al. [54]. This cell architecture does not require a thin transparent
conductive oxide (TCO) layer on a transparent glass ceramic superstrate, and is a cost
effective and robust manufacturing process for thin-film cells. Other studies indicate
that an effective operation to improve light trapping properties of the solar cell is to
directly texturise the TCO that has previously been deposited on the glass superstrate
[52, 55, 56]. One possibility of texturing glass substrates is with the use of laser
ablation mechanisms.
2.13 Silicon quantum dot solar cells
Much work has been done in recent research on quantum dot solar cells. There are
many potential applications for quantum dots, from multi-junction tandem band-gap
tuned architectures to photon energy down-converters [7, 57]. Silicon quantum dots in
particular have been subject to much research due to their optical properties and
biocompatibility [58]. It has been shown in literature that it is possible to change the
optical properties of silicon quantum dots, not only by controlling their size, but also
by controlling their surface chemistry [58-65]. The band-gap of the quantum dots can
be controlled by controlling the diameter of the dot. As the size of the dot is reduced
below the exciton Bohr radius in a semiconductor, the quantum confinement effect
occurs, which causes significant changes to the optical properties compared to those
found in the same bulk material [7]. By reducing the diameter of the quantum dot it is
possible to increase the band-gap of the material, thus offering the possibility to
optimally tune the band-gap of the material to suit the wavelength of light incident
upon it.
There are several possible architectures in which quantum dot based solar cells can be
realised. One such design, the quantum dot sensitised solar cell, shows particular
promise [7, 66, 67]. This type of design is based on the sensitisation of a micro-to-
26
nano porous TiO2 paste deposited on the surface of a transparent conducting substrate.
It has been reported that the porosity, and pore size, and the pore connectivity are
important issues for the efficient operation of this type of solar cell [66]. The benefits
of solar cells based on TiO2 are well known. Due to the simplicity of the cell design
manufacturing costs can be reduced, and furthermore, there are optical benefits in the
use of TiO2 films as the micro and nano porous structures of these materials greatly
enhance light scattering [68]. The benefits of this include less sensitivity to
illumination directionality, and the possibility of enhanced light trapping effects. The
high surface area of the nanostructured materials also enable the efficient absorption
of light [69].
Figure 11: Example of a quantum dot sensitised solar cell architecture
[70].
27
2.14 Solar Cell superstrate material selection
When choosing a superstrate material for this investigation there were several key
factors that were taken into account, as follows.
refractive indices
laser absorption characteristics
thermal and physical properties and stabilities
cost and availability
transparency to the AM1.5G spectrum
2.14.1 Refractive indices
In order to ascertain the feasibility of the laser texturing of the front cover superstrate
of a solar module it is necessary to choose a superstrate material with similar optical
properties to those used in conventional solar modules produced today. The majority
of solar modules today use low-cost soda lime glasses, which tend to have refractive
indices of between 1.50 and 1.55 [71]. However, it is possible that more expensive
glasses such as fused silica may be used in cases where maximum power output is of
highest importance. Refractive index is an influential parameter of this study also, as
it will affect the amount of deflection the light rays will undergo during the
geometrical light trapping effect induced by the textured surfaces.
2.14.2 Laser absorption characteristics
As most glasses are transparent across the visible electromagnetic spectrum, it is
important to choose the materials carefully as how they absorb laser light is highly
influential in determining the process quality and ability of the glass using a laser
process. Unless the laser used is able to deliver pulses of light short enough with a
high enough peak intensity to cause non-linear and two-photon absorption, such a
picosecond or femtosecond laser, it is unlikely that a laser operating in the visible
spectrum will be able to successfully obtain the required level of energy absorption
for micro-machining [39, 72]. Far-infrared lasers are well suited to the machining of
glasses as the laser light at this wavelength is absorbed well by SiO2 structures [39,
73, 74].
28
2.14.3 Cost and availability
Soda-lime is the most commonly used glass, and is widely used in applications ranging
from window glass to solar module cover glass due to its low cost and wide availability
[71]. Other glasses such as borosilicate are higher in cost, and are used in instances
where the physical properties of soda-lime glass are not sufficient. Borosilicate glass,
widely known as Schott glass or Pyrex, is widely used in cookware for instance due
to its excellent thermal characteristics [75]. Meanwhile, fused silica is a synthetically
manufactured glass used only the highest quality base materials and is of high purity.
This enables fused silica to exhibit its high transmission, high chemical resistance, and
excellent thermal shock characteristics [72, 76]. While fused silica is significantly
more expensive than it counterparts, its properties can be quite beneficial for certain
applications, such as optics for high power lasers and windows in highly sensitive
scientific equipment.
2.14.4 Thermal properties and stability
While most optical glasses are brittle, they exhibit a high level of hardness and
excellent chemical resistance [1]. The thermal conductivities of glasses is normally
quite poor. While poor thermal conductivity can mean that it is possible to laser
machine locally with a very small heat affected zone (HAZ), it can also mean that the
material is very sensitive to thermal shock and cracking [73]. Fused silica is known to
be one of the glasses with the highest thermal-shock resistance of all common optical
glasses, making is highly suitable for laser micromachining. Having this high thermal-
shock resistance means it is possible to laser machine the material with a long-pulse
infra-red laser, where thermal breakdown is the mechanism for laser ablation during
micromachining.
2.14.5 Transparency to AM1.5G spectrum
According to the ASTM G173-03 Reference Spectra, the AM1.5G spectrum ranges in
wavelength from the ultra-violet range at 280nm to the infrared at 4000nm [1].
However, as the electronic band-gap of silicon is 1.1eV, silicon will not absorb any
light with a wavelength longer than approximately 1100nm [1, 77, 78]. Photon
energies close to the band-gap of the silicon absorber also provide the highest spectral
response of the solar cell, therefore it is most important that the superstrate material is
29
transparent in UV-VIS-IR region of the spectrum. Higher energy photons, with
energies greater than 1.1eV, contribute the most significant energy flux of the
spectrum and results in the majority of the carrier generation that occurs in the solar
cell. Most commercially available glasses are highly transparent between 350nm and
1100nm [52].
2.15 Literature Review Summary
From this past research, it is clear that there remains a lot of interest and potential
scope for solar cell superstrate topology optimisation. In this work fused quartz was
chosen as the solar cell superstrate material to be investigated. This material provides
favourable thermal shock characteristics and excellent optical transmission properties
across the UV-VIS-IR spectrum. In particular, in this work a detailed design of
experiments for the laser texturisation of this superstrate material was investigated for
its effect on solar cell efficiency. Two characterisation methods were also developed
and investigated to allow for more detailed characterisation of the solar cell structures.
These included the design, commissioning, and verification of an optical profilometer
based on the depth-from-focus principle for the metrology of textured superstrates. In
this work an analysis of the FIB-SEM serial sectioning method was completed, which
has applications in the evaluation of macro-porous solar cell materials.
30
Chapter 3 Materials & Methods
3.1 Design of a Depth-from-Focus based Optical Profilometer
The main investigation of this work was to assess the effect of laser texturised glass
superstrates on the performance of solar cells. In order to thoroughly evaluate how
solar radiation interacts with a textured surface and subsequently with the solar cell
behind it, it is necessary to measure the surface shape and texture characteristics
accurately. The laser processing parameters can then be correlated with the measured
surface texture parameters. These surface characteristics of the solar cell can in turn
be related to the cell performance under various conditions.
Optical profiling systems offer capabilities that are well aligned to this research, as
they can measure with high-resolutions over larger surface profile depths compared to
stylus type instrumentation. The non-contact nature of the measurement is also
beneficial, as the optical profiling solutions can perform a measurement over a large
surface area within a short measurement time. Examples of widely used optical
profiling techniques for transparent and optical materials include White-Light
Interferometry (WLI) and Laser Scanning Confocal Microscopy (LSCM). While these
techniques are widely utilised in the measurement of optically smooth surfaces, they
have limitations to their capabilities when measuring optically rough surfaces,
especially due to effects from surfaces roughness with a high spatial frequency [79,
80]. While the term ‘optically rough’ is somewhat arbitrary and is highly dependent
on the end application of the optical surface, in literature it has been referred to as
surface roughness in the range of 10 nm to 100 µm Ra when considering WLI
measurement applications [79]. The surface profiles created by laser processing glass
can vary in depth and roughness, with some channels exhibiting steep flanks in excess
of 70° [81]. These characteristics can be challenging to measure accurately using
optical techniques, as the numerical aperture of the objective lens used in the
measurements is directly correlated to the maximum surface angle that can be
measured [79]. In WLI applications for example, these factors can result in data
dropout, as the interference signal becomes too weak to provide any level of certainty
at particular points of the measurement. This data then requires interpolation, which
can introduce unknown amounts of measurement error.
31
In pursuit of the above research goals it was decided to perform an initial evaluation
of a 3D non-contact optical profilometer based on the depth-from-focus principle, a
method which requires varying the focal plane of a specially designed optical and
imaging system with a low depth-of-field, from which it is then possible to extract
surface contour information from the surface of the specimen [82]. Advantages of this
method include sub-micrometre level resolution and the ability to measure and
visualise surfaces in real colour. This measurement technique has shown good promise
in the measurement of steep flanks and varying surface roughness conditions [80].
Little research has been performed on the suitability of an instrument with this
operating principle for the measurement of optically rough transparent surfaces,
although the measurement of translucent materials has been reported as a limitation of
this technique [80]. As part of this work an optical profilometry system was designed
and fabricated (see Figure 12), which facilitated the initial evaluation of the
measurement technique with the aim of testing the feasibility of the depth-from-focus
method for the measurement of optically rough transparent surfaces. Part of the device
evaluation process included the manufacture of a brass measurement artefact by
milling, which was evaluated using other measurement instruments (WLI and stylus)
with two main aims: firstly, to provide a method of initial comparison of the designed
profilometer with commercially available systems, and secondly, to provide an
approximate dimensional verification of the designed system. While the traceable
calibration and evaluation of such an artefact requires significant amount of work,
testing, and statistical analysis with a certified metrology institution (such as NIST,
NPL or PTB), the artefact manufactured in this work provides some initial information
about the general accuracy of the device and also provides a method by which the
measurement repeatability can be evaluated.
As the systems contains components of motion control, image processing, surface
display, and signal analysis, NI LabVIEW was selected as the software development
environment due to its ability to integrate with third party hardware and the availability
of additional software libraries such as NI Vision.
32
CMOS Camera
X-axis Stepper Motor
Y-axis Stepper Motor
Z-axis Servo Motor
Granite base
LED and XY Motor Controller
50x Microscope Objective
LED Inline Lighting
Figure 12. 3D non-contact optical surface profilometer.
33
Figure 13: Main Graphical User Interface (GUI VI Front Panel) for the 3D non-
contact optical surface profilometer.
Figure 14. Flowchart indicating operation of the 3D non-contact optical surface
profilometer control software.
34
3.1.1 System Specification
Table 3. Optical system specification
Objective lens model Nikon CFI60 LU Plan Epi ELWD Infinity
Corrected
Objective lens magnification 50x
Objective lens working distance 10.1 mm
Objective numerical aperture 0.55
Objective resolving power 0.5 µm
Objective depth-of-field 1.0 µm
Lens tube model Infinitube FM-100
Lens tube focal length 100 mm (0.5x)
System magnification 25x
System field-of-view 310 µm x 232 µm
Camera model Mightex Systems MCE-C030-U, USB2.0
CMOS camera, 3MP colour, Enclosed, w/
GPIOs, C-mount
Camera sensor Aptina ½" Micron MT9T001, colour, Rolling
Shutter
Camera pixel size 3.2 x 3.2 µm
Camera resolution 2048 x 1536 pixels (1024 x 768 with 1:2
decimation)
Image sampling distance 0.303 µm per pixel
Illumination Single white LED, potentiometer adjustable
output, co-axial illumination configuration
35
Table 4. Motion control system specification
Z-axis
Actuator model Thorlabs 612B
Motor type 12 V DC with 256:1 gearing
Encoder type 48 count encoder
12,288 counts/rev with gearing
Leadscrew pitch 0.5 mm
Travel range 12 mm
Resolution 40 nm
Speed 50 – 425 µm/s
Load capacity (axial) 4.5 kg
Controller model Precision MicroControl Corp. DC2-PC 100
Series Motion Controller
X & Y axes
Motor model Nanotec L4018S1204-M6
Motor type Bi-polar stepper, 3.5 V, 1.2 A
Steps/revolution 200
Leadscrew pitch 1 mm
Resolution 5 µm full step, 2.5 µm half-step
Speed 0.005 – 5 mm/s
Controller model Byvac BV4113
3.1.2 Design of the Optical System
The high resolution optical system was implemented using a 50x magnification Nikon
microscope objective lens. This objective lens was diffraction limited and has the
power to resolve features in the order of 500 nm in size using white light illumination
while providing a very shallow depth-of-field of 1 µm (as specified by the
manufacturer), meaning only a fraction of the image is focused at any one time. The
depth-from-focus measurement technique takes advantage of the limited depth-of-
field to correlate various height points on a surface with a vertical scanning position.
As the limited depth-of-field is the basis of the measurement method, using an
objective with a larger depth-of-field (with a lower numerical aperture and
36
magnification) would reduce the vertical measurement resolution of the device. The
objective lens was used in conjunction with a 100 mm focal length microscope tube
lens and a 3MP colour USB CMOS camera as the imaging device. This configuration
gave the optical system an overall effective magnification of 25x with a field of view
of 310 µm by 232 µm, as determined in section 3.2 using a calibrated stage
micrometer.
3.1.3 Design of the Motion Control System
There are two elements to the motion control system (Figure 15). The vertical
positioning system was used to scan the sample along the vertical axis of the optical
system, thus changing the position of the in-focus optical plane. A high-accuracy
Thorlabs servo actuator (612B), with a stated minimum incremental movement
capability of 200 nm, was used for the vertical positioning of the specimen and was
controlled using a Precision Micro Control (DC2-PC) servo motor driver/amplifier.
Third party hardware integration is possible as LabVIEW supports many third party
hardware components. The vertical positioning system had a theoretical resolution of
40nm. The XY positioning system was controlled using two Nanotec stepper motors
(L4018S1204-M6) using third party USB stepper drivers (Byvac BV4113), providing
a theoretical resolution of 5µm when full-stepping, allowing for the positioning of the
specimen and for the implementation of mosaic acquisition with image stitching for
large area measurement.
a) b)
Figure 15. Images of the (a) high resolution optical system and vertical
actuator, and (b) XY positioning stage with stepper motors.
37
3.1.4 Image acquisition and processing
The raw data acquired in the form of a stack of images is then processed to extract the
focused regions from each image in the stack. As the vertical position of each image
is known from the synchronisation of the motion control system during the image
acquisition, it is possible to build a 3D surface mesh from the extracted focused
regions. Significant work has been done in the area of image processing algorithms
for the extraction of depth data [82-103]. There are a significant number of algorithms
available which offer various advantages for various applications. In this
implementation, it was decided to utilise a straightforward focus detection method
through the use of a gradient magnitude algorithm. The feature extraction was
performed through the use of the Sobel edge detection algorithm in this instance [104].
The Sobel algorithm is essentially an edge detection algorithm. By using a Sobel edge
detector, an intensity value is assigned for the gradient of every pixel in the image
being analysed. The stack of images was then parsed to find the highest gradient
intensity in the image stack for each pixel along the vertical measurement plane. The
highest intensity was the referenced to a height according to the vertical position that
image was recorded at, which is stored in a ‘Height Map’. This process is shown in
Figure 16.
The 3D surface can then be visualised in full colour by placing the focused colour
image data onto the height map. An inbuilt LabVIEW VI was used to process the
height map of the surface, which is then able to display in a 3D Picture Control. Using
a 3D Picture Control configuration VI it is possible to wrap the fully focused image
(generated previously from the stack of images by extracting the focused regions) on
the surface mesh, allowing the measured specimen to be displayed in full colour at
sub-micron resolution as shown in Figure 17.
38
Figure 16. Raw data processing procedure. Both a topographical height
map and a high resolution fully focused image are created.
Figure 17. 3D surface display of laser processed grooves in transparent
glass. The surface has been shaded blue to improve visibility.
Measurement area is 310 µm by 232 µm.
39
3.1.5 Image stitching (tiling)
When measuring large areas using a high-magnification optical configuration, it is
often necessary to take measurements at various positions on the sample which are
later combined together to create a larger image at high resolution. This process is
frequently referred to as ‘tiling’, but is also known as image stitching. There are
various approaches to this method, however typically this is achieved by overlapping
the tiled areas of the acquisition. Overlapping the tiled sections allows for the
registration of the measurement areas in all three axes of the co-ordinate system. A
typical overlap amount is 10% which is generally seen as a reasonable compromise
between data integrity and measurement speed [79]. In the implementation of image
stitching in this work, no overlap or image registration algorithm was utilised in this
initial investigation. Instead, tiled areas were placed beside one another in the correct
order to form a larger high-resolution image.
3.1.6 Surface roughness calculation and evaluation
It is often necessary to measure the surface roughness of components in surface
engineering, as the surface roughness can significantly affect the performance of a
surface in a particular application. One example of this is in the manufacture of high
tolerance mechanical parts, such as internal combustions engine components.
However, with relevance to solar cells, surface roughness is a useful evaluation
parameter as this can influence light scattering properties of a particular surface [105].
Different surface roughness parameters reveal different important characteristics of
the material and the production processes. However, there are strict guidelines that
must be met in order to make roughness analysis repeatable and accurate. Firstly, the
data must be corrected for form error by linear regression. The following step involves
using a Gaussian filter with a low cut-off frequency defined by ISO 4287, which
removes the waviness component of the surface [106]. The waviness may appear as a
repeating pattern in the data, whose wavelength is too large to be considered as
roughness. A visual description of the components of surface roughness are given in
Figure 18.
40
Figure 18: (a) A representation of the actual surface being measured. (b)
The components of the surface profile according to ISO 4287.
In order to evaluate during the calculation exactly what surface form error and
waviness constitute, it is necessary to consult ISO 4287 and ISO 4288 [106, 107]. This
standard specifically defines the data filtering parameters that must be adhered with in
order to achieve a compliant surface roughness measurement. These parameters are
shown in Table 5.
41
Table 5: Selection of Filter parameters according to ISO 4288
Periodic
profiles Non-periodic profiles
Cut-off
lengths
Sampling
length Evaluation length
Spacing distance (mm) Rz (mm) Ra (mm) lc (mm) l (mm) L (mm)
> 0,01 to 0,04 to 0,1 to 0,02 0,08 0,08 0,4
> 0,04 to 0,13 > 0,1 to 0,5 > 0,02 to 0,1 0,25 0,25 1,25
> 0,13 to 0,4 > 0,5 to 10 > 0,1 to 2 0,8 0,8 4
> 0,4 to 1,3 > 10 to 50 > 2 to 10 2,5 2,5 12,5
> 1,3 to 4 > 50 > 10 8 8 40
The remaining filtered data is then used for the roughness analysis (Figure 19). Two
types of analysis were programmed and performed using the developed software; line
analysis and area analysis, where the area analysis provides similar information to that
of the line analysis but in two directions instead of one.
Figure 19. Roughness profile is obtained using the ISO 4288 standard for
profile correction.
42
3.2 Verification of the optical plane of the 3D Profilometer
The optical XY plane was verified using an optical stage micrometer (PYSER-SGI
PS78) which had been calibrated and certified by UKAS. The measurement
uncertainty using this calibrated micrometre was ±0.0005 mm, which allowed for
accurate adjustment of the optical system. Figure 20 shows an image of the calibrated
micrometre scale taken as a resolution of 2048 pixels by 1536 pixels with a 1:2 image
decimation ratio, which meant the recorded image has an effective resolution of 1024
pixels by 768 pixels. By measuring the number of pixels in the image against the
calibrated scale it was possible to calculate the camera calibration value to be 3.3
pixel/μm, the reciprocal of which gives 0.303 μm/pixel. Using this information, it is
possible to calculate that the field-of-view of the optical imaging system at this
resolution was 310 μm by 232 μm. The adjustment of the system was achieved by
storing the 0.303 μm/pixel value in the operating software of the system. This value
was then taken into account when calculating the scale of the final data output.
Figure 20 - PYSER-SGI PS78 Stage micrometer used for calibrating the
optical plane of the microscope system.
43
3.3 Performance analysis of the linear stages of the 3D Optical Profilometer
In order to assess the performance of the linear stages, it was necessary to utilise
measurement tools to evaluate the positional performance characteristics of the
devices. The aim of the performance analysis was to determine the positioning
accuracy and repeatability of the low-cost linear stage configurations for use in a 3D
optical profilometry applications, data which can subsequently be used when
evaluating the overall system measurement performance. To assess the accuracy and
repeatability, it was necessary to follow a standard procedure to ensure that any values
obtained for the stage performance was in line with international standards and that
they could be compared readily with the performance of other devices.
The international standard ISO 230-2 standard is specifically designed for the analysis
of stage performance [108]. The standard describes the methods which should be
followed, and also details the selection of parameters used in testing. An extract from
the standard of the parameters and formulae used in this analysis are included in
Appendix D.
3.3.1 Selection of Target Positions for Positional Analysis
The analysis was performed by moving the linear stages point to point, from one target
position to the next. The stages were allowed a settling time of a minimum of 5 seconds
for every measurement taken, in order to ensure the axes had come to a complete stop,
and that and residual vibrations did not affect the position measurement.
The X and Y stepper motor driver axes and the Z servo motor driven axis were driven
point to point at a velocity of 1mm/s with a linear acceleration profile of 10mm/s2.
These low velocities and accelerations were chosen specifically in order to negate the
effect of static friction and inertia on the accuracy and repeatability of the system.
The value of each target position can be selected without restrictions. However, it
should take the general form presented in (1), in order ensure any period errors within
the motion system are correctly sampled.
44
𝑃𝑖 = (𝑖 − 1) 𝑝 + 𝑟 (1)
Where i is the number of the current target position; p is an interval based on a uniform
spacing of target points over the measurement travel; and r is a random number within
± the amplitude of possible periodic errors in the movement mechanism. It is
recommended by the standard to use a changing value of this random number, for each
target position, between ±30 % of P, which allows for any periodic errors to be
adequately sampled and evaluated.
3.3.2 Recording of the stage positions using Dial Gauge Indicators
In order to evaluate the positional accuracy and repeatability of the linear stages, dial
gauge indicators were used to measure point to point movements. In this approach, a
dual scale analysis was performed. The reason behind the dual scale approach is that
some Dial Gauge indicators facilitate longer travels, which those that have shorter
operating ranges typically offer improved accuracies. By performing the dual scale
analysis, it allows the linear stage to be evaluated for accuracy and repeatability for
both large and small scale movements. In this work, the two Dial Gauges utilised were
the Mahr MarCator 810SW and the Mitutoyo 513-405E for the long and short range
analyses respectively. Details of the analysis are outline in Table 6 and Table 7.
Table 6: Analysis of large scale movements using Mahr 810SW dial
gauge
Axis Analysis Nominal
Range (mm)
Gauge
Measurement
Range (mm)
Measurement
Point Spacing
(mm) (p)
X-axis 5 10 0.5
Y-axis 5 10 0.5
Z-axis 0.5 10 0.05
45
Table 7: Analysis of small scale movements using Mitutoyo 513-405E
dial gauge
Axis Analysis Nominal
Range (mm)
Gauge
Measurement
Range (mm)
Measurement
Point Spacing
(mm) (p)
X-axis 0.2 0.2 0.02
Y-axis 0.2 0.2 0.02
Z-axis 0.2 0.2 0.02
Measurement performed for with the Mahr Dial Gauge were recorded to the nearest
0.01mm, as this was the resolution of the dial. For measurements performed with the
Mitutoyo Dial gauge, measurements were recorded to the nearest 0.001mm.
Specification sheet for both devices are available in Appendix C. The measurement
setup for the axes is shown in Figure 21. In both cases, the gauges were carefully
aligned perpendicular to the stage motion using an engineer’s square tool in order to
minimise the influence of cosine error on the measurement.
Figure 21: Mitutoyo 513-405E Dial Gauge when performing point to
point measurements on the X-axis of the 3D Optical Profilometer.
In accordance with recommendations of ISO 230-2, the target positions for the point-
to-point measurement evaluation were chosen according to (1), where p is the nominal
46
measurement interval, and r is the random variation from that position. The random
position was listed as a percentage of the target position P with values as listed in
Table 8. This ensured the proper sampling of any periodic errors in the system.
Table 8: Position variation from the nominal point-to-point interval.
Target Positions (i) 1 2 3 4 5 6 7 8 9 10
Random Position
Variation % (r) -5 9 2 -12 10 -4 -12 -7 10 2
In order to proceed with a repeatable experimental procedure, and to comply with ISO
230-2, the point-to-point sequence as shown in Figure 22 was followed. Starting at the
top left of the figure, the stage movements move from the initial position to the first
target position, P1. Subsequent movements are then made the remaining points in the
cycle. It is important to note that while make the 3 individual cycles, the measurement
setup is not reset to a zero position. This procedure then ensures that any cumulative
errors are sampled across the three cycles.
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
Cycle 1
Cycle 2
Cycle 3
Target Positions (Pi)
Cycle j
Figure 22: The measurement procedure was performed according to the
following sequence.
The actual target positions used in the analysis for each axis are shown in Table 9 and
Table 10 for the large movement and small movement analyses respectively. In Table
10 the values for the X and Y target positions have been rounded to the nearest 5 μm,
and that is the theoretical minimum resolution of the stepper motor driven stages. The
Z-axis has a smaller travel range than the X and Y axes, hence a smaller range of
motion is investigated for the Large Movement analysis, as shown in Table 9.
47
Table 9: Target positions for Large Movement analysis. All positions in
millimetres.
Axis P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 END
X 0 0.48 1.09 1.53 1.76 2.75 2.88 3.08 3.72 4.95 5.10 5.5
Y 0 0.48 1.09 1.53 1.76 2.75 2.88 3.08 3.72 4.95 5.10 5.5
Z 0 0.05 0.11 0.15 0.18 0.28 0.29 0.31 0.37 0.50 0.51 0.55
Table 10: Target positions for Small Movement analysis. All positions in
millimetres.
Axis P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 END
X 0 0.020 0.045 0.060 0.070 0.110 0.115 0.125 0.150 0.200 0.205 0.25
Y 0 0.020 0.045 0.060 0.070 0.110 0.115 0.125 0.150 0.200 0.205 0.25
Z 0 0.019 0.044 0.061 0.070 0.110 0.115 0.123 0.149 0.198 0.204 0.22
3.3.3 Potential sources of measurement uncertainty
In evaluating the stage positioning the potential sources of measurement error must be
considered. There are five main sources of error that must be considered in this work.
Gauge accuracy
The stage position measurements are limited by the resolution of the dial gauge. The
limiting factor in term of accuracy is the accuracy of the gauge over its full length of
travel. In the case of the Mitutoyo 513-405E, the measurement uncertainty is 3 µm
over the range of travel, while the Mahr 810SW has an uncertainty of 17 µm over 10
mm of travel.
Cosine error
Cosine error occurs when the axis of a measurement instrument becomes misaligned
with the axis of motion [109]. For an angular misalignment of 10°, a measurement
compensation factor of 0.98 must be applied to the measurement result (as cos(10) =
0.98). For example, over a 0.2 mm measurement, a 10° misalignment would mean the
total measurement deviation due to cosine effects would be 4 µm. However, over a 5
48
mm range, the cosine error would be in excess of 0.1mm, which is significant. In this
work the gauges were set up using an engineer’s square tool to minimise the cosine
effect.
Abbe error
Abbe error occurs when the measurement plane of a sample changes its angular
attitude as it traverses an axis [109]. This type of error can be caused by inherent
wobble in a motion stage, which is caused by the stage not being perfectly straight or
flat. This in turn causes the stage to pitch and roll, an effect that can be greatly
exaggerated if the sample being moved is far from the motion stage. To evaluate the
type of error complex optical interferometry equipment is required. In order to
minimise the influence of Abbe error in this work, the gauges were placed as close as
practically possible to the normal position of an artefact during measurement.
Thermal effects
Thermal effects cause materials to expand and contract, which can affect not only the
device being measured, but also the measurement gauge itself. The dial gauges are
calibrated for used at 20°C according to the manufacturers specifications. No
temperature controlled environment was available for use during this work, however
the stage position measurements were recorded in quick succession (within 1 hour for
each stage) in order to minimise any thermal effects. For example, the thermal
expansion of an aluminium stage would be on the order of 25 µm/m.K, which for a
stage 50 mm in length equates to a change in length of 1.25 µm per Kelvin.
Vibration effects
Vibration from external sources can cause measurement variations, and can cause
particular issues when performing dynamic position measurements (i.e. where the
stage is measured while in motion). In this work, the designed profilometer was rigidly
mounted to a granite base to lower the centre-of-gravity of the system and to reduce
vibration effects. During the measurement, the stage was allowed to come to a
complete stop at each measurement point with a dwell time of 5 seconds. There were
no visible effects of vibration influencing dial gauge readings as each measurement
was recorded.
49
3.4 Verification of 3D measurement performance
In order to assess the 3D measurement performance of the 3D profilometer, it was
necessary to utilise a 3D measurement artefact. However, the majority of
commercially available 3D measurement standards that are calibrated accurate to the
1 micrometre range are specifically designed for either for Atomic Force Microscopes
(AFM) or While Light Interferometers (WLI). The key issue around this lies in the
fact that these standards are normally made from flat polished materials, such as
silicon wafers [110]. While these extremely flat polished surfaces work well in the
calibration of AFM and WLI systems, they are not ideally suited to optical
microscopy, as imaging using this technique depends entirely on the ability to form a
high-contrast image of an object. A surface with a minimum average roughness of 15
nm and a spatial period of 2 µm are required in order for the sample to be measurable
using standard brightfield microscopy [80]. Little work has been performed in the area
of the calibration of 3D profilometers utilising the depth-from-focus measurement
technique that is in line with international ISO standards. Sun and Claverley utilised
high grade tungsten carbide spheres in the range of 1.5mm to 3mm, with the aim of
being able to acquire reliable calibration and verification measurement data within one
field-of-view of the commercial measurement system [111]. This approach utilised the
ISO 10360-8:2013 standard, which was specifically designed for the calibration of
micro-co-ordinate measurement machines. The conclusion of the work was that
different sampling patterns across the field-of-view resulted in different levels of
measurement error, which measurement errors of up to 20 micrometres found when
utilising a 50x microscope objective.
In order to test the step-height measurement performance of the 3D optical
profilometer, a purpose machined measurement artefact was manufactured. This block
was manufactured from brass material utilising a 0.75mm end mill tool as shown in
Figure 23.
50
Figure 23: SEM image of a 0.75 mm milling tool utilised for
manufacturing the measurement artefact.
The brass block was machined to leave long regular columns protruding at the top
surface. Each column was 160 micrometres in width and 100 micrometres in step
height, subject to general machining tolerances. An oblique image of the measurement
artefact is presented in Figure 24.
Figure 24 - Wide field Backscatter Electron SEM image of machined
brass sample
51
Figure 25 - Secondary Electron SEM image of end face of a pillar on the machined
brass sample. Machining debris is visible at the end face of the measurement
artefact.
In order to initially verify the dimensions of the columns, an SEM view of the end face
of the measurement artefact was obtained, as shown in Figure 25. The step height of
the sample was measured to be between 103 micrometres and 105 micrometres for all
columns. While this measurement acts as a good verification check, it cannot be
utilised for the purposes of measuring and calibrating the sample over an area. In order
to validate the brass artefact, it was necessary to take measurements of the same
sample utilising other measurement equipment that had been adjusted using other
certified calibrated reference artefacts. It must be mentioned that to produce a certified
calibrated artefact requires a significant amount of investment, and was outside the
main scope of this work. The main aim of this work was to verify that the performance
of the developed measurement system was in an expected range in order to show that
the measurement technique could be used for micrometre scale measurement of
optically rough transparent superstrates. For this level of verification, it was acceptable
to utilise other available laboratory measurement systems in order to gain an initial
understanding of the capabilities of the developed optical profilometry device,
52
however for the evaluation of the absolute measurement accuracy of the system
significantly more investigation and statistical analysis of the artefact would be
required. In this instance, the verification equipment consisted of a Bruker DEKTAK
V200Si Stylus Profilometer, and a Veeco WYKO NT1100 White Light
Interferometer, with both instruments are sensitive well into the sub-100nm
measurement range. In an effort to maintain measurement consistency, all
measurements taken in this analysis were taken at the same point on the measurement
artefact, the location of which was marked on the sample. The measurement taken
with these White Light Interferometer is shown in Figure 26 and Figure 27.
Figure 26 - 3D surface measurement of the machined brass sample using
a Wyko NT1100 white light interferometer
53
Figure 27 - Cross sectional Wyko NT1100 measurement data of
machined brass sample. Step height is measured as 103µm.
The measurement data shown in Figure 27 indicates the step height at the desired
measurement location to be 103µm. The step height measurement taken at the desired
measurement location using the stylus profilometer is shown in 3D in Figure 28 and a
2D cross-section is shown in Figure 29. It can clearly be seen in these Figures that the
edge definition of the brass artefact column is lost when measuring with the stylus
profilometer, as might be expected. Overall, there was a very close correlation between
the results from the SEM, WLI, and stylus profilometer in measuring the step height
of the brass artefact, especially considering the preliminary nature of this
investigation. The results output from the measurement instruments are presented in
Table 11.
Table 11: Measurement results of brass sample step height taken by the
various instruments
Instrument SEM WLI Stylus
Result in µm 104 103 102
54
Figure 28 - 3D surface profile of the machined brass sample using a
DEKTAK V200Si Stylus profilometer
Figure 29 - Cross sectional profile of machined brass samples acquired
using DEKTAK V200Si Stylus profilometer
55
3.4.1 Testing procedure of the 3D measurement performance of the Optical
Profilometer
In order to evaluate the 3D measurement performance of the Optical Profilometer, a
dual scale measurement approach was taken. A multifactorial Design of Experiments
was utilised to evaluate the performance over measurement ranges of 103 µm and 512
µm respectively. For the 103 µm measurement, the purpose machined brass block
described in section 3.4 was utilised. For the 512 µm measurement, a nominal 500 µm
gauge block from a Mitutoyo Rectangular Gauge Block Set (as shown in Figure 30)
was used. This gauge block was first verified by WLI, and the results of this are shown
in Figure 31.
Figure 30: Mitutoyo Rectangular Gauge Block Set (Grade 1)
56
Figure 31: Step height of nominal 500µm Mitutoyo Rectangular Gauge
Block. The WLI measurement verified the dimension of the block to be
512µm.
To evaluate the 3D measurement performance of the 3D Optical Profilometer, a testing
procedure was designed in the absence of a formal ISO standard for the evaluation of
this type of instrument. Accuracy and repeatability can be determined by direct
comparison of measurement of a sample with a known dimension. However, as a
certified calibrated artefact was not available, this became a challenging aspect. By
utilising the purpose machined brass artefact and the Mitutoyo gauge block as the step-
height references it is only possible to verify the estimated accuracy of the system as
the absolute accuracy of the artefacts is not certified or traceable. The estimated
accuracy is sufficient for an initial evaluation of the system, but for absolute accuracy
verification further work would be required. The accuracy is evaluated by taking
nominal step height to be the average of the three measurements, as detailed in Table
11. While this approach does not give absolute accuracy information, it is designed to
give some credibility to the investigation in measuring at this scale. The repeatability
of the measurements is an aspect that can be evaluated, as this doesn’t depend on the
absolute accuracy of the artefact. Having a repeatable measurement can be considered
more important, as the accuracy of the system can easily be improved at a later date
providing the system is repeatable.
57
To evaluate the 3D measurement performance of the system the following procedure
was formulated. Firstly, the most important measurement parameters had to be
calculated. In this instance, the most important controllable measurement parameter
was sampling height on the Z-axis. As outlined in section 3.1.4, the profilometer
system operates by recording a series of images along the optical axis of the
instrument, which is parallel with the Z-axis. Hence, by controlling the physical
distance between each image recorded in that sequence, it is possible to evaluate the
3D measurement accuracy as a function of sampling interval. This correlation is of
interest from a practical point of view, as it is vital to know the limitations of such a
measurement device. By having larger sampling intervals, it is equally possible to
increase the speed of acquisition and measurement, as less images would need to be
recorded over a given measurement height and less images would need to be post-
processed to produce the results. The sampling intervals explored in this investigation
are outlined in Table 12. Each measurement was taken at a similar location along the
Z-axis in order to maintain consistency between the measurements. Also, all
measurements were taken at the same locations on the sample. Five repetitions of
measurements with the same sampling interval were also performed to assess the
repeatability of the measurements. Each measurement in Table 12 was performed 6
times in a row, which also gives some data around the short-term repeatability of a
measurement.
58
Table 12: Design of Experiments with specified sampling intervals
Run
Reference step
height
(μm)
Sampling
Interval
(μm)
1 512 0.6
2 512 1
3 103 0.6
4 103 0.6
5 103 0.6
6 103 0.2
7 103 1
8 512 0.2
9 103 0.6
10 103 0.6
In order to remove extraneous noise from the results, for example, due to areas with
steep flanks, the resulting height maps were post processed to remove unwanted noise.
An example of this is seen in Figure 32, which is a measurement of the brass
measurement artefact recorded with the developed system. The areas shown as white
in Figure 32 (d) are removed from the results, as the aim in this work is to evaluate the
step-height via the upper and lower plane locations only. This is a similar approach to
that used by Sun and Claverley, as 95% of measurement points are selected for
evaluation with the removal of outliers [111]. The procedure for fitting planes to the
data involved the selection of the upper and lower planes in the height map, calculating
the average height of the plane, and then subtracting Lower Plane Height from the
Upper Plane Height. The data for these calculations are available in Appendix G. The
accuracy is then calculated based on how close the average step height measurement
is to those of the estimated artefacts, while the standard deviation of the measurements
provides data about the repeatability of the measurements.
59
Figure 32: a) 3D view of measurement, b) Fully focused image from a
step height measurement, c) the height map from this measurement, d)
the areas of high un-repeatability and noise due to steep flanks.
60
3.4.2 Potential sources of measurement uncertainty in the 3D measurement
There are several potential sources of measurement uncertainty when considering the
3D measurement of surfaces. The following is not an exhaustive list of potential
sources, but are considered to be the most significant in this investigation.
Z-axis accuracy
As the vertical position of any point on the surface to be measured is correlated to the
on-axis position of the Z-axis, the accuracy of the Z-axis is critical to the overall
accuracy of the measurement system. In this work the Z-axis accuracy is evaluated of
small measurement distances to gain an understanding of the how the Z-axis influence
the overall 3D measurement performance of the system.
Principal optical aberrations
There aberrations include spherical, comatic, chromatic, astigmatism, field curvature,
and distortion. The influence of these aberrations on the measurement can be greatly
reduced by using high-quality optics. The Nikon CFI60 LU Plan objective utilised in
the developed system is well corrected for chromatic, spherical, and field curative
aberrations. More details on these types of aberrations can be found in literature [80,
112].
Spatial resolution
The minimum resolvable distance between two point sources is known as the Rayleigh
criterion [80]. In the developed profilometer, the optical resolution of the objective
lens is the limiting factor for spatial resolution, which is 0.5 µm using the 50x Nikon
objective with a numerical aperture of 0.55. The limiting of the spatial resolution
determines the limit of the spatial roughness frequencies that can be resolved by such
an optical system.
61
3.5 CO2 Laser System hardware and specifications
The laser chosen for the texturisation of the glass surface was a Rofin DC-015 CO2
laser operating at 10.6 µm wavelength. This laser was chosen as it operates in the
infra-red (IR) spectrum, which is absorbed well by the fused quartz glass. As this
wavelength is also well absorbed by carbon based materials these lasers are well suited
to cutting polymers and other organic materials such as wood [39]. These lasers are
also available with high continuous powers (up to several kW) making them ideal for
the gas-assisted cutting of thick materials and metals. This particular model laser also
offers operation in pulsed and continuous wave modes, allowing for fine process
tuning and control. According to the supplier specifications, the laser power stability
to +/- 5% over the operating range of the laser, which may be attributed to the fact the
laser cavity and active medium is diffusion cooled through the use of water cooled
electrodes, allowing for better output stabilities due to more consistent temperatures
[113]. This laser was integrated onto a CNC motion control system built by Mectronic
Systems Ltd. and is capable of machining 2D profiles from materials at velocities up
to 1m/min.
Figure 33. Graphic illustration of pulse period, duration and peak power
within pulse width.
In order to control the output power of the laser, a modulation signal is fed from the
CNC controller to the laser, the duty cycle of which is proportional to the instantaneous
speed of the motion axes. This allows for greater control of the laser machining process
during period where the motion axes undergo acceleration.
Pulse Period
Pulse Duration
Time (s)
Pe
ak P
ow
er
(W)
62
The percentage duty cycle was calculated as follows using (2).
𝐷𝑢𝑡𝑦 𝐶𝑦𝑐𝑙𝑒 % =
𝑃𝑢𝑙𝑠𝑒 𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛
𝑃𝑢𝑙𝑠𝑒 𝑃𝑒𝑟𝑖𝑜𝑑× 100 (2)
Table 13. Laser parameters held constant during the experiments
Parameter Value
Peak power 375 W
Assist gas and pressure Argon at 1.5 bar
Nozzle standoff 5 mm
Focal position On the top surface
Focal length of lens 127 mm
Beam diameter at focal point Approx. 200 µm at 1/e2
3.6 Glass materials selection for solar cells superstrates
Following the consideration of the material characteristics required for this
investigation, the decision was made to proceed with fused quartz material for the
solar cell superstrates. This material provides favourable thermal shock characteristics
and very good transmission properties across the UV-VIS-IR spectrum. The material
was supplied by Multi-Lab Ltd. (Newcastle-upon-Tyne, UK), and was ordered in
dimensions of 25 mm by 25 mm squares with thicknesses of 1 mm and 2 mm. The
datasheet including properties and transmission spectrum for this material is available
in Appendix E.
3.7 Design of Experiments
To carry out an analysis of the effects of laser processing parameters on the parasitic
absorption of light (losses) in laser processed glass a Design of Experiments (DOE)
was implemented. Through a process of parameter screening, it was observed that the
63
most influential process parameters were laser irradiance (W/mm2), pulse repetition
frequency (Hz), and traverse scanning speed (mm/s). A 33 central composite factorial
design of experiments was then planned using levels determined during the screening
process (Figure 34). The average power of the laser, and thus the irradiance of the
beam, was altered by holding the peak power of the laser constant and adjusting the
duty cycle of the pulses.
Figure 34. Central Composite Design of Experiments
Table 14. Laser process parameters used for DOE
Duty Cycle (%) PRF (Hz) Speed (mm/s)
Level 1 2 160 1.67
Level 2 4 280 5.00
Level 3 6 400 8.33
64
The processes used for each sample in the DOE are presented in Table 15. There
parameters represent the combination of all the parameters listed in Table 14 that were
used on each particular sample with made of the 33 multifactorial DOE. The column
labelled Average Power (W) is a calculated parameter from the Duty Cycle % value
listed and the Laser Peak Power setting defined in Table 13.
Table 15: Laser micromachining parameters for each sample produced in
the DOE.
Channel
#
Duty Cycle
(%)
Average
Power (W) PRF (Hz)
Speed
(mm/s)
1 2.00% 7.5 160 1.67
2 2.00% 7.5 276 1.67
3 2.00% 7.5 400 1.67
4 4.00% 15 160 1.67
5 4.00% 15 276 1.67
6 4.00% 15 400 1.67
7 6.00% 22.5 160 1.67
8 6.00% 22.5 276 1.67
9 6.00% 22.5 400 1.67
10 2.00% 7.5 160 5.00
11 2.00% 7.5 276 5.00
12 2.00% 7.5 400 5.00
13 4.00% 15 160 5.00
14 4.00% 15 276 5.00
15 4.00% 15 400 5.00
16 6.00% 22.5 160 5.00
17 6.00% 22.5 276 5.00
18 6.00% 22.5 400 5.00
19 2.00% 7.5 160 8.33
20 2.00% 7.5 276 8.33
21 2.00% 7.5 400 8.33
22 4.00% 15 160 8.33
23 4.00% 15 276 8.33
24 4.00% 15 400 8.33
25 6.00% 22.5 160 8.33
26 6.00% 22.5 276 8.33
27 6.00% 22.5 400 8.33
65
3.8 ANOVA Analysis
During the creation of the Response Surface Model (RSM) an Analysis of Variance
(ANOVA) was also carried out to determine how well the model data fitted to the
actual process values, and a description of the Response Surface Methodology can be
found in Appendix B. This statistical analysis was performed using Design Expert
V8.0. In order to evaluate the experimental data, a quadratic model was fitted to the
data points in the design space shown in Figure 34. This allowed for the analysis of
the linear interactions and quadratic effect of each of the parameters used in the Design
of Experiments. The fitted quadratic model was represented by the following form as
shown in (3).
√% 𝐴𝑏𝑠𝑜𝑟𝑝. = 𝑏0 + 𝑏1𝑥1 + 𝑏2𝑥2 + 𝑏3𝑥3 + 𝑏4𝑥1𝑥2 + 𝑏5𝑥1𝑥3 +
𝑏6𝑥2𝑥3 + 𝑏7𝑥12 + 𝑏8𝑥2
2 + 𝑏9𝑥32
(3)
where 𝑥1 represents the Pulse Repetition Frequency (Hz), 𝑥2represents the Traverse
Speed (mm/s), and 𝑥3 represents the Duty Cycle (%).
3.9 Etching of processed glass with Hydrofluoric Acid (HF)
In order to further reduce the amount of parasitic light absorption in the laser textured
glass, it was necessary to post-process the surface using a chemical etching technique.
Hydrofluoric acid is widely used in electronics for the removal of SiO2. In order to
assess the effect of the chemical etching on the surface, 9 samples of the 27 (numbers
19-27) were immersed in HF (40%) for a period of 5 minutes. Following etching they
were rinsed for 3 periods of 5 minutes in de-ionised water. The integrating sphere
setup described in section 3.10 was then used to measure any change in the light
absorption characteristics of the textured glass due to acid etching.
66
3.10 Integrating sphere setup for reflection / absorption / transmission
measurements
In order to measure the effect of the laser processing on light absorption in the fused
quartz glass, an integrating sphere with spectrometer setup was designed. Integrating
spheres have the ability to collect diffuse and specular reflections and transmissions
from samples when irradiated with light [114]. As these specialised spheres are coated
with highly reflective diffuse coatings, it is possible to average any transmission or
reflection over the whole sphere. This makes it possible to obtain an accurate average
of any transmitted or reflected light, regardless of directionality of the light rays. The
integrating sphere (Labsphere model 4P-GPS-040-SF) used in this setup was 100 mm
in diameter with three 25.4 mm diameter ports at 0°, 90° and north pole locations, and
a 38 mm port at 180°. A baffle was located between the 90° and 180° ports to reduce
measurement error due to light falling directly on the detector from the sample. To
further improve the accuracy of the measurement a cosine corrector was utilised,
which aids the elimination of any directionality from the measurement. The
spectrometer (Ocean Optics model USB650-VIS-NIR Red Tide) had a spectral
resolution of 1nm and a spectral measurement range from 360 nm to 1000 nm,
however, due to weak UV illumination provided by a broadband light source a
significant amount of noise could occur until approximately 420 nm. Data below this
wavelength are not utilised.
Figure 35. Integrating sphere setup for absorption measurements.
67
Figure 36. Cut-away view of the integrating sphere setup
To accurately measure the light absorption in the laser processed fused quartz, the
samples were held inside the closed sphere with the illumination source incident at 8°.
This allowed all the scattered light to be collected including transmitted and reflected,
diffuse and specular. By measuring the amount of light collected in comparison to that
of a reference measurement with clear unprocessed fused quartz, the absorption could
be calculated using the following equation (4).
𝐴𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 = 1 − (𝑇 + 𝑅) (4)
where T is the transmitted light and R is the reflected light.
3.11 Correcting absorption data to AM1.5G solar irradiance spectrum
In order to evaluate the absorption in the glass for solar cell applications the measured
absorption values were transposed with solar spectral energy distribution data. The
measured data was corrected using spectral data from the AM1.5G standard using the
following method.
68
1. The measured data had a spectral range from 420nm to 1000nm, however the
irradiance data obtained spanned a larger spectrum from 280nm to 4000nm [14].
The data was presented as W/m2/nm and it was first necessary to convert the
irradiance data to a percentage of the measurement spectrum using (5).
% 𝑆𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝑖𝑟𝑟𝑎𝑑𝑖𝑎𝑛𝑐𝑒𝜆 =
𝐼𝜆
𝐼𝑇× 100 (5)
where Iλ is the irradiance at a given wavelength, and IT is the total solar irradiance
between 420nm and 1000nm of the AM1.5G spectrum.
2. The data from the previous operation was then multiplied by the co-efficient of
absorption found during the initial measurement, as shown in (6).
% 𝐴𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 𝑜𝑓 𝐴𝑀1.5𝐺 𝑠𝑝𝑒𝑐𝑡𝑟𝑢𝑚
=% 𝑆𝑝𝑒𝑐𝑡𝑟𝑎𝑙𝑖𝑟𝑟𝑎𝑑𝑖𝑎𝑛𝑐𝑒𝜆
× % 𝐴𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛𝜆 (6)
3.12 SEM and EDXS Analysis
A Zeiss EVO LS15 scanning electron microscope (SEM) was utilised to visualise the
surface topographies of the laser processed samples. This device was also used to
perform energy dispersive x-ray spectroscopy analyses on the processed surfaces
using an INCA detection and analysis system supplied by Oxford Instruments. Ten
random points were selected for compositional analysis of the sample surfaces. The
EDXS parameters used to perform this analysis are shown in Table 16.
Table 16. EDXS compositional analysis parameters
Parameter Value
Accelerating voltage 10kV
Probe Current 1nA
Working Distance 8.5mm
Magnification x1000
69
3.13 Channel Dimensional Analysis
To assess the different dimensions of channels created by laser micromachining it was
necessary view the cross section of the channels. This was achieved by laser cutting
the material to reveal a cross–section of the channel, and them imaging of the cross-
section by SEM as described in section 3.12. This approach proved effective in the
analysis of the dimensions, as the cross-sections reveal that some of the machined
channels have a large aspect ratio, making depth measurements difficult using other
measurement methods. The cross-sections were performed in three random locations
along the channel in order to achieve an average depth, taking into account variations
that might occur in the depth due to the nature of the process.
Three measurement were recorded at each inspection locations, as shown in Figure 36
and Figure 38. The “Depth” was measured from the top surface of the material to the
visible bottom of the channel cross-section. The “Width 50%” was recorded at 50%
depth in the channel cross-section. In order to take repeatable measurements of the
“Width Top” dimension, it was necessary to employ a procedure by which the location
of the channel opening was placed. In this instance an imaginary reference line was
drawn across the surface of the material, and the “Width Top” measurement locations
were placed at the point where the channel opens up below the reference line, as is
shown in Figure 38. All measurements were recorded to the nearest micrometre.
Dep
th
Width 50%
Figure 37: Diagram showing the SEM measurement locations on a cross
sectional view of laser micro-machined channel.
70
Width Top
Figure 38: Diagram showing the SEM measurement locations for the
measurement of the Width (top) parameter. Top surface is represented by
red line, and measurement locations are shown in orange.
3.14 Solar cell selection and preparation
Commercially available solar cells (Evergreen Solar, String Ribbon™ wafers) were
diced into eight by eight millimetre sections using a 355 nm nanosecond pulsed laser
source (Blueacre Technology Ltd.). The front and rear contacts of the solar cells were
soldered manually. To provide refractive index matching media in the void between
the solar cell and the superstrate, the solar cells were encapsulated in an elastic
transparent resin (ACC Silicones, QSil 215) and allowed to set at room temperature
for 24 hours.
3.15 Measurements of I-V vs angle of incidence
The set-up for I-V curve measurement of the solar cell is shown in Figure 39 and
Figure 40. This measurement was performed for samples 20, 21, and an untextured
reference as superstrates at varying angles of incidence between a normal incidence
angle (90°) and an incidence angle of 10° in increments of 10°. The textured glass
superstrate and the encapsulated solar cell were held in contact using the device shown
in Figure 40 whereby securing screws allowed for the application of sufficient pressure
in order to eliminate air gaps at the glass/encapsulant interface. The photovoltaic
measurements (I-V characteristic) of the solar cell in conjunction with different
textured superstrates were carried out using a Newport Oriel solar simulator. The solar
simulator was equipped with a 450W Xe lamp filtered to AM1.5G spectral conditions
and an illumination area of 78 cm2. The light intensity was adjusted for using a NREL-
71
calibrated reference silicon cell. The data was recorded using an automated data
acquisition system (Gamry PCL4 300 series) under 870W/m2 solar simulated
irradiance and an externally applied bias potential. The backside of the cell was
carefully shielded to avoid any influence of stray light.
Figure 39. a) Angle of incidence of 90°, and b) angle of incidence of 10°.
Figure 40. Mounting device constructed to hold textured superstrate and
encapsulated solar cell in contact. The rotational stage allowed for
measurements of varying angles of incidence to be performed.
72
3.16 Optical characterisation of alkyl-capped silicon quantum dots
The silicon quantum dots utilised in this work were produced by collaborators using a
procedure already reported [59]. An outline of the procedure and parameters utilised
in the preparations of this material is provided below [115].
3.16.1 Preparation
Luminescent silicon nanocrystals (quantum dots) were produced by electrochemical
etching of silicon. Porous silicon (PS) with a nanoscale internal structure was
produced by a standard etching procedure from silicon. The starting point was to
remove the native oxide layer from a silicon chip by dipping it in HF for a short time,
washing with de-ionized water and blowing with a stream of nitrogen. The chip was
then clamped between two O-rings in a Teflon cell of diameter 6 or 10 mm, which
contains a HF–ethanol (electrolyte) mixture in the ratio 1:1. Due to the hydrophobic
character of the clean Si surface, absolute ethanol is usually added to the aqueous
solution to increase the wettability of the silicon surface. In fact, ethanoic solutions
infiltrate the pores, while purely aqueous HF solutions do not. This is very important
for the lateral homogeneity and the uniformity of the porous silicon layer in depth.
Another important reason for adding ethanol is that during the reaction hydrogen
evolution occurs, bubbles are formed and stick on the Si surface in pure aqueous
solutions, whereas they are promptly removed if ethanol is added. The electrical
contact was obtained via thin copper wire in the base of the Teflon cell. A coiled
tungsten wire (0.5 mm) was used as the counter electrode. Silicon nanoparticles were
obtained after 15 min etching at current density of (380-400 mA /cm2). The electrolyte
was removed from the cell while the nanocrystals remained on the chip. These
particles were dried in a Schlenk flask under vacuum. The dried nanocrystals were
introduced into a flask containing a mixture of the 2 ml of toluene or mesitylene and
0.1 ml of undecenol solution. The solution was refluxed for 2-6 hrs under nitrogen
atmosphere. During this refluxing process, the whole internal structure of the
nanocrystals is modified. This process is termed alkylation. During this reflux only the
smallest silicon nanoparticles broke off the porous silicon matrices and dissolved in
the solution. The particles have been called silicon nanocrystals due to their size (3-5
nm in diameter).
73
3.16.2 Silicon quantum dots size evaluations
In order to validate the size of the silicon quantum dots, our collaborators in this
project performed some additional investigations. The material was cast onto a mica
substrate and characterised by Atomic Force Microscopy (tapping mode), as shown in
Figure 41 [115]. The results indicate that some individual quantum dots are on the
order of 5 nm in diameter, there are also clusters of quantum dots which form after
subsequent depositions on the substrate, with diameters of up to 30 nm to 40 nm found,
which can be attributed to inhomogeneous evaporation of the solvent. A previous
study by the same group on the surface topography by AFM on the similarly produced
sample indicated a uniform nanocrystal size distribution of between 3-5 nm [116, 117].
STEM analysis on the same sample has also been previously performed to verify the
dimensions of these silicon nanocrystals [117].
74
-1 0 1 2 3 4 5 6 7
0
1 0
2 0
3 0
4 0
5 0
He
igh
t (n
m)
la te r a l p o s it io n ( m )
a)
b)
Figure 41: The AFM image was taken on the Si-QDs cast onto mica. From the
profile along the line (a), one can find out the height of clusters was about 10 nm,
which is equivalent to the height of two layers of Si-QDs, as shown in (b) [115].
75
3.16.3 Deposition of silicon quantum dots onto textured glass superstrates
For optical absorption characterisation fused quartz glass was used as a superstrate.
This 1 mm thick glass was laser cut into 10 mm by 10 mm square sections and cleaned
using Piranha solution (3:1 mixture, concentrated H2SO4, 30% H2O2) at 90°C for 25
minutes. The Si-QD was dispersed in dicholoromethane [118] and then deposited
using the drop deposition method. 4 ml of the solution was dropped per deposition,
and the substrate allowed dry for 24 hours in a clean environment to allow for complete
evaporation of the solvent.
To evaluate the initial absorption characteristics of the material, three depositions were
performed in total for measurement made on untextured fused quartz superstrates.
To evaluate the increase in absorption in the Silicon quantum dots when used in
conjunction with laser textured superstrates, five depositions were performed in total
for measurement made on both untextured and textured fused quartz superstrates. The
textured superstrates used in this investigation were subject to the same laser
parameters and HF etching post treatment as outline in section 3.7 and section 3.9
respectively.
It is important to note that in this measurement the absorption results have not been
weighted against the AM1.5G solar spectrum, as the aim of this study was purely to
investigate the absorption characteristics of the Si quantum dots relative to air. This
was to allow proper correlation of the absorption spectrum with the band-gap of the
material, and also to facilitate comparison of the results with those reported in
literature.
76
3.17 Focused Ion Beam Serial Sectioning
Silica monolithic rods (MonoTrap RCC18) used in this work were obtained from GL
Sciences (Tokyo, Japan) and used as received. The carbon monoliths utilised in this
work were produced by collaborators using a procedure already reported [119-122].
An outline of the procedure and parameters utilised in the preparations of this material
is provided below.
3.17.1 Chemicals
1-Butanol and ferric chloride (99 %) were obtained from Riedel-De Haen (Seelze,
Germany). Resorcinol (99 %), formaldehyde (37 wt % in water) and concentrated
hydrofluoric acid (HF, 38-40 %) were obtained from Sigma-Aldrich (Gillingham,
UK). Silica beads with a diameter of 5 μm (surface area, 94 m² g-1; pore size, 15 nm)
were used as templates for the preparation of the carbon monolithic rods. Deionised
water (18.2 MΩ∙cm) was obtained from a Millipore Direct-Q™ 5 water purification
system (Millipore, Bedford, MA, USA).
3.17.2 Preparation of carbon monolith
The carbon monolith was prepared according to a procedure adapted from that
reported by Eltmimi et al. [123]. A 1 g portion of 5 μm silica particles were dispersed
in 1.5 g of 1-butanol under sonication for 1 h. This was followed by the addition of
0.18 g of ferric chloride and 0.367 g of resorcinol. A gentle shaking was applied to
dissolve the resulting mixture and form the resorcinol/Fe complex. Then, 0.3 g of ice-
cooled, 37% formaldehyde solution in water was added to the mixture and stirred for
1 h in an ice-water bath. The resultant mixture was polymerised in 6 mm ID glass
tubes kept at 90°C for 16 h in a water bath (GFL, model 1013, Laborgerateborse
GmbH, Burladingen, Germany). The resulting phenolic resin/silica rods were then
removed from the glass tubes and aged for 72 h in a fume hood to slowly evaporate
the majority of the residual solvent. Finally, the rods were dried under vacuum at 80°C
overnight using an EHRET thermovacuum oven (Ehret Labor and Pharmatechnik
GmbH, KG, Emmendingen, Germany), and further cured at 135°C for 4 h to ensure
complete polymerisation.
A desktop alumina tube furnace (model GSL1300X, MTI, Richmond, VA, USA)
purged with N2 was then used to pyrolise the rods. Initially, the temperature was
ramped from room temperature to 800 °C at a rate of 2.5 °C min-1, and held at this
77
temperature for 2 h. A second ramp was applied from 800 °C to 1250 °C at a rate of
10 °C min-1, keeping this temperature for a further 1 h. Then, the furnace was allowed
to cool down to room temperature Next, the silica particles and the iron catalyst were
removed from the rods by etching in 38% HF for 5 h and a subsequent wash with
copious amounts of deionised water. Finally, the carbon monolithic rods were
thoroughly dried under vacuum at 80 °C for 16 h.
3.17.3 Image acquisition
A small portion of the carbon and silica monolithic rods were subjected to FIB-SEM
analysis with a FEI Quanta 3D FEG DualBeam instrument (FEI Ltd, Hillsboro, USA)
integrating a Ga+ liquid metal ion source (spot size < 5 nm). Prior to the FIB-SEM
analysis, the samples were fixed onto the stub using silver paste and subsequently
coated with a thin layer of gold (thickness, ~ 7 nm) using a Emitech K575X Sputter
Coating Unit (Quorum Technologies, UK). FIB sectioning was carried out at an
acceleration voltage of 30 kV to etch a section of approximately 60 x 50 µm to a depth
of 100 nm for the carbon monolith (current, 0.3 nA), and of 55 x 40 µm to a depth of
66.7 nm for the silica monolith (current, 0.5 nA). A 2D image of each section was
automatically acquired by the SEM at a magnification of 2000x, an acceleration
voltage of 5 kV, and a nominal current of 5.92 pA for the carbon monolith and of 1.5
pA for the silica monolith. This FIB-SEM sequence was automatically repeated 100
times for the carbon monolith and 150 times for the silica monolith in a set routine. A
total of 100 images were thus collected for the carbon monolith, each one being 100
nm further into the structure compared to the preceding image. Likewise, a total of
150 images were collected for the silica monolith, each one being 66.7 nm further into
the structure. The angle of image acquisition was 52° in relation to the direction of the
FIB as shown in Figure 42. The volumes and dimensions of the acquisition are
presented in Table 17 and Table 18.
78
Table 17: Acquisition details for carbon monolith
Original Volume 0.000045262 mm3
Analysed Volume 0.000015119 mm3
Slice thickness 100 nm
Pixel dimension 74 nm
Number of Slices 93
Dimensions of Analysed volume 58.24 μm x 26.49 μm x 9.30 μm
Table 18: Acquisition details for silica monolith
Original Volume 0.000049073 mm3
Analysed Volume 0.000014732 mm3
Slice thickness 66 nm
Pixel dimension 74 nm
Number of Slices 150
Dimensions of Analysed volume 44.69 μm x 33.74 μm x 9.90 μm
Figure 42: Sketch of the sample/FIB-SEM system set-up for serial
sectioning and imaging.
79
3.17.4 Image processing
The SEM images resulting from the FIB serial sectioning procedure (Figure 43 and
Figure 44) were processed via image analysis routines to extract the pore boundaries
and reconstruct the 3D structure. First, the images were resized anisotropically to
correct for the image acquisition angle of 52°, after which the images were subjected
to an image registration processes where landmarks were identified in the sequence of
images. These landmarks were then used to calculate the necessary translation
required to minimise any image alignment error. This was achieved using the formula
for a rigid body translation (7),
𝒙 = {(cos 𝜃 , − sin 𝜃), (sin 𝜃 , cos 𝜃 ) } ∙ 𝒖 + ∆𝒖 (7)
where u and x are the input and output vectors of the landmarks, respectively, ∆u is a
constant vector that gives the amount of translation, and θ is the angle of rotation in
the image sequence.
Figure 43 Image captured by SEM after FIB sectioning of the carbon
monolith.
80
Figure 44 Image captured by SEM after FIB sectioning of the silica
monolith.
The resulting images were then cropped to reduce the area of analysis to approximately
58.2 μm x 26.5 μm (width x height) for the carbon monolith and 44.7 μm x 33.7 μm
for the silica monolith, with a resolution of 74 nm/pixel. From the images captured for
the carbon monolith, only the first 93 were used for analysis. Thus, the depth of the
final image stack was 9.3 μm, with a resolution of 100 nm/pixel in the depth direction.
The depth of the final image stack employed for the reconstruction of the silica
monolith was 10 μm, with a resolution of 66.7 nm/pixel in the depth direction.
No attempt was made to increase the resolution of the captured images by increasing
the SEM magnification, thus decreasing the size of the detectable pores. The scope of
this work was simply to demonstrate the suitability of FIB-SEM analysis for
reconstruction of the bulk porous structure in conducting and non-conducting
monolithic materials, as have applications in chemistry and in next generation solar
cells. Therefore, a rather large reconstructed volume was considered to be more
representative of the overall monolith morphology, as compared to a reconstructed
volume just a few cubic micrometres in size that would have been needed for
visualisation of the nanoporous structure.
81
Finally, the two image stacks were binarised after manual outlining and filling of the
porous structures, as shown in Figure 45.
Figure 45 From L to R: Outline of the pores, filling of pores and
binarisation applied to the image stack obtained for the carbon monolith.
3.17.5 3D reconstruction and morphological analysis
Following binarisation, a 3D histogram of each of the image stacks was
produced, enabling the calculation of the total macroporosity of the volume under
analysis by (8).
% 𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦 =
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑣𝑜𝑥𝑒𝑙𝑠 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑖𝑛𝑔 𝑝𝑜𝑟𝑒𝑠
𝑇𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑣𝑜𝑥𝑒𝑙𝑠 𝑖𝑛 𝑣𝑜𝑙𝑢𝑚𝑒 × 100 (8)
To assess the accuracy of the resulting macroporosity values, Mercury Intrusion
Porosimetry (MIP) measurements were also carried out with the carbon and silica
monoliths by collaborators [122]. MIP is an established porosity measurement
technique used in the characterisation of open pore materials with pore size ranges
from 3.5 nm up to 500 μm [124]. The total porosity and pore size distribution were
determined with an Autopore IV 9500 Series (Micromeritics Instrument Corporation).
Dried samples with weights ranging from 36 mg to 47 mg were used for these
measurements.
The 3D reconstruction of both monoliths from their respective image stacks was then
performed using the volume rendering technique available within the Materialise
Mimics 13.0 software (Materialise NV, Belgium). This allow for the 3D visualisation
of the porous structure and allows for a much better understanding of the connectivity
of the pores within the material.
82
Chapter 4
Verification of the 3D non-contact
optical profilometer
4.1 Positional Analysis of the XYZ stage system
In order to assess the quality of the low-cost 3D imaging system it was necessary to
measure the accuracy and repeatability of the motion systems used during data
acquisition. The microscope had three automated axes: X and Y axes to position the
sample in the optical plane, and the Z axis to position the sample along the optical axis
to vary the focal position.
The X and Y axes by design were constructed in a manner to reduce the cost of the
positioning system. The purpose of the X and Y axes were twofold; firstly, the axes
would be used to position the sample accurately for data acquisition, and secondly the
axes would be used for the automated acquisition of large area measurements through
the use of area stitching. The theoretical positioning resolution was designed to be 5
μm through the use of a low-cost stepper motor / thread-bar leadscrew arrangement,
which consisted of a 200 steps/revolution stepper motor combined with a 1mm pitch
leadscrew. Theoretically this would mean that the XY stage could position the sample
to the nearest 5 μm (16 pixels in the image), which from an area stitching perspective
was sufficient as additional data processing for accurate area stitching would be to
allow for accurate combination of acquired images. There is also the option of micro-
stepping the stepper motors which can greatly increase positioning resolution,
however for the purposes of this study full-step stepper motor drivers were utilised.
The Z axis forms a critical part of the accuracy of the 3D measurement capabilities of
the system. The control software was developed such that images were recorded at
varying user defined focal positions along the optical axis. However the exact values
of those positions are constrained by the accuracy and repeatability of the Z axis.
Hence it was necessary to assess the performance of this axis in order to evaluate the
overall potential of the system as a low-cost 3D metrology solution. The axis utilised
for the Z axis motion was selected on the basis of several important engineering
83
considerations, namely compact size, resolution, minimum incremental movement
performance, and cost.
The following analyses utilises the ISO 230-2 standard to assess the motion
performance of these axes. Two performance characterisation tests were completed; a
large scale movement analysis test and a small scale movement analysis test. This dual
scale approach was utilised in order to measure the performance of the low cost linear
stages for 3D profilometry applications. The large scale analysis was performed with
a Mahr MarCator 810SW dial gauge indicator with a measurement accuracy of ±17
μm over the 10 mm range of travel. The small scale analysis was performed with a
Mitutoyo 513-405E dial gauge with an accuracy of ±3 μm over the 0.2 mm range of
travel.
4.1.1 Large scale positional analysis
The target positions and stage movement results for the large scale analysis for the X,
Y, and Z axes are presented in Figure 46, Figure 47, and Figure 48 respectively. The
calculated repeatability and accuracy results for these stages are presented in Table 19,
Table 20, and Table 21 respectively.
Further results involved in the calculation of the stage accuracy and repeatability from
this investigation are presented in Appendix A.
84
Figure 46: Large scale positional analysis along X axis.
85
Figure 47: Large scale positional analysis along Y axis.
86
Figure 48: Large scale positional analysis along Z axis.
87
Table 19: Results from large scale positional analysis along X axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 10
Mean Reverse value �̅� NA NA 1
Range mean bidirectional
positional deviation 𝑴 NA NA 40
Systematic positional
deviation 𝑬 30 30 40
Repeatability 𝑹 40 40 80
Accuracy 𝑨 60 60 60
Table 20: Results from large scale positional analysis along Y axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 20
Mean Reverse value �̅� NA NA 2
Range mean bidirectional
positional deviation 𝑴 NA NA
10
Systematic positional
deviation 𝑬
10 20 20
Repeatability 𝑹 80 40 140
Accuracy 𝑨 80 60 80
Table 21: Results from large scale positional analysis along Z axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 10
Mean Reverse value �̅� NA NA 1
Range mean bidirectional
positional deviation 𝑴 NA NA
20
Systematic positional
deviation 𝑬
20 10 20
Repeatability 𝑹 40 40 80
Accuracy 𝑨 60 50 60
88
The results from the large scale positional analysis indicate that, as expected, the
highest bi-directional performance was achieved by the servo driven Z-axis, closely
followed in terms of performance by the X-axis. According to this analysis the Z-axis
operates with similar repeatability in both directions when moving uni-directionally.
When the 3D non-contact optical profilometer is performing a measurement, the Z-
axis only moves uni-directionally during the measurement. Hence it could be argued
that the uni-directional is the most important performance factor for the Z-axis. While
this axis provided the best performance of the three axes, it should be noted that the Z
axis was tested over a smaller travel range than the X and Y axes, due to the limitation
in travel displacement available from the Z-axis.
The X and Y axes are both identical in design, differing only in the mass loaded on
the stages. The X axis supports the sample platform and the Z-axis actuator, while the
Y axis supports the same load in addition to the mass of the X axis. From an
operational perspective, the bi-directional performance of the X and Y axes are
important, as if a large area 3D measurement is being performed then the positioning
repeatability of the sample in the XY plane of the profilometer is of importance. The
X and Y axes achieve almost identical performance when considering the results
calculated from the Unidirectional – movements. In the Unidirectional + movements,
the difference in the accuracies reported is within the limit of error of the Mahr
MarCator 810SW dial gauge indicator. This indicates that the X and Y axes operate at
similar performance levels over the nominal 5 mm range investigated. The most
significant sources of positioning error for the X and Y axes can be attributed to
manufacturing defects in the leadscrews, inaccuracy in the step positions of the stepper
motors, and the accuracy limitations of the measurement device.
4.1.2 Small scale positional analysis
The target positions and stage movement results for the small scale analysis for the X,
Y, and Z axes are presented in Figure 49, Figure 50, and Figure 51 respectively. The
calculated repeatability and accuracy results for these stages are presented in Table 22,
Table 23, and Table 24 respectively.
Further results involved in the calculation of the stage accuracy and repeatability from
this investigation are presented in Appendix A.
89
Figure 49: Small scale positional analysis along X axis.
90
Figure 50: Small scale positional analysis along Y axis.
91
Figure 51: Small scale positional analysis along Z axis.
92
Table 22: Results from small scale positional analysis along X axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 2
Mean Reverse value �̅� NA NA 0
Range mean bidirectional
positional deviation 𝑴 NA NA 4
Systematic positional
deviation 𝑬 4 3 4
Repeatability 𝑹 4 8 14
Accuracy 𝑨 8 11 8
Table 23: Results from small scale positional analysis along Y axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 5
Mean Reverse value �̅� NA NA 1
Range mean bidirectional
positional deviation 𝑴 NA NA 4
Systematic positional
deviation 𝑬 5 5 7
Repeatability 𝑹 12 24 40
Accuracy 𝑨 15 24 15
Table 24: Results from small scale positional analysis along Z axis
Axis Deviation (all in µm) Unidirectional + Unidirectional - Bidirectional
Reverse value 𝑩 NA NA 11
Mean Reverse value �̅� NA NA 1
Range mean bidirectional
positional deviation 𝑴 NA NA 2
Systematic positional
deviation 𝑬 2 3 12
Repeatability 𝑹 8 12 22
Accuracy 𝑨 10 13 11
93
The small scale positional analysis revealed some interesting results. Over these small
scale movements, the stepper motor driven X-axis outperformed the servo motor
driven Z-axis in terms of accuracy and repeatability, both uni-directionally and bi-
directionally. The high performance of the X-axis bi-directionally can be ascribed to
the fact the stage is directly driven by the motor, with minimised leadscrew backlash.
However, the Z-axis operates with a small servo driven motor through a planetary
gearbox. While gearboxes are an efficient means of transferring power, they can also
introduce backlash into the motion system. Nevertheless, it is important to note that
the X-axis set-up would not be a suitable replacement for the current Z-axis, as in
order to record 3D measurements using a small depth with the depth-from-focus
principle, the ability to move in repeatable small increments is required. In the current
configuration the X-axis is limited to a positioning resolutions of 5 μm, which is too
course a sampling interval for the measurement of micro-scale surface profiles.
The X-axis also outperformed the Y-axis in terms of accuracy and repeatability, which
can be attributed to the fact that the Y-axis motor must drive a higher mass. The
difference in performance between the two axes could potentially also be attributable
to other factors, such as manufacturing variation between the respective leadscrews or
motors.
The results from this investigation revealed that the stage performance in the small
scale analysis is superior to the stage performance in the large scale analysis. For
example, the X-axis bi-directional accuracy improves from 60 μm for the large scale
displacement to only 8 μm for the small scale displacement, however one must also
take into consideration that the former measurement is over a longer distance.
94
4.1.3 Verification of 3D measurement performance
For each verification experimental run listed in Table 25, the step height measurements
were repeated six times. The complete list of displacement measurement results from
these runs are presented in Appendix G.
Table 25: Accuracy results from the 6 sub-run measurements
Run Reference step
height (μm)
Sampling
Interval
(μm)
Avg.
Height
(μm)
Accuracy
% Std Dev (μm)
1 512 0.6 511.6 99.92 0.9
2 512 1 503.4 98.32 0.4
3 103 0.6 103.2 99.81 0.2
4 103 0.6 103.1 99.90 0.1
5 103 0.6 102.7 99.71 0.4
6 103 0.2 102.4 99.42 0.1
7 103 1 102.5 99.51 0.2
8 512 0.2 503.3 98.30 0.7
9 103 0.6 102.9 99.90 0.2
10 103 0.6 102.6 99.61 0.1
As can be seen from the results in Table 25, the accuracy for measurements taken with
the two measurement artefacts was always in excess of 98%. While this is not
confirmation of absolute accuracy, it does give some confidence that the system can
measure effectively at the 100 µm step height and the 500 µm step height scales. The
standard deviations across the run measurements are all below 1 μm, which offers a
good reflection on the repeatability of the step height measurement process using the
developed 3D non-contact optical profilometer. It can also be noted that the highest
repeatability (lowest standard deviation) results were achieved when measuring over
103 µm, while the 512 µm measurement scale returned the lowest repeatability results.
This contrasts well with the fact that the Z-axis manufacturer’s specification indicates
a minimum incremental movement of 0.2 μm. When sampling at 0.2 μm over the 103
μm step height, the results recorded a high degree of repeatability, however when the
95
same sampling interval is utilised over the 512 μm step height measurement, the
repeatability is reduced. It could be argued that this small sampling interval is
repeatable over distances of 100 μm, however due to manufacturing and operational
variations in the Z-axis, that sampling interval is not as repeatable over a distance of
500 μm. It is also important to note that the Z axis verification measurements
performed in section 4.1.2 are relative to an arbitrary zero point, and the repeatability
of the stage is evaluated over several run with respect to that same fixed point in space.
However, in the 3D measurement it is only the relative distance between the upper and
lower plane positions that is evaluated for repeatability.
4.1.4 Comparison of measurement results with a commercially available system
In order to qualitatively evaluate the performance of the developed 3D non-contact
optical profilometer in comparison with a commercially available system, a
measurement of the machined brass sample was evaluated using a Keyence VHX-
5000 series microscope. This system is based on a proprietary “depth-from-defocus”
principle, and thus offers a good performance comparison with the system developed
in this work. The measurement of the brass artefact by the Keyence VHX system is
shown in Figure 52, and is a combination of two 1 mm x 1 mm measurements stitched
together.
Figure 52: Measurement recorded using a Keyence VHX-5000 series 3D
microscope.
96
The measurement taken with the 3D non-contact optical profilometer that was
developed in this thesis is shown in Figure 53 and Figure 54, the first of which displays
the fully focused full colour image output from the developed software, and the latter
shows the 3D measured profile of the step-height. It can clearly be seen that the
developed system preserves the edge sharpness of the machined brass step, as the steep
flanks to the side of the column are more accurately represented in the measurement
in Figure 54 compared with the measurement shown in Figure 52. The 3D profile
shown in Figure 54 visually compares well with the cross sectional SEM image shown
in Figure 25. While this is only a qualitative evaluation, it would be fair to say that the
developed system performs at least comparatively with the commercially available
Keyence system.
Figure 53: Fully focused true colour image of the machined brass sample
97
Figure 54: 3D representation of the surface profile of the machined brass
sample.
4.1.5 Demonstration of capability in the measurement of translucent materials
The 3D profile shown in Figure 55 was recorded using the 3D non-contact optical
profilometer developed in this thesis, and demonstrates the profilometer capabilities
in measuring translucent laser micro-machined surfaces. The measurement area is
approximately 1.2 mm by 0.9 mm, and is composed by the automatic stitching of 16
(four by four) 3D profile measurements.
Figure 55: 3D measurement profile of laser micro-machined grooves in
fused quartz glass. Axis values are in nm.
98
Chapter 5 Laser texturisation of
glass superstrate for enhanced
solar cell performance
5.1 Laser processing and parasitic light absorption losses
Significant alterations in the parasitic absorption loss characteristics of the glass due
to different laser processing parameters were observed, as is shown in Figure 56(a)
and Figure 56(b). Figure 56(a) shows the influence of traverse speed (mm/s) with the
PRF and Duty Cycle fixed at levels of 400 Hz and 6% respectively. It can be seen that
the light absorption is inversely related to the traverse speed. As the traverse speed
increases the absorption of the processed surface decreases. Figure 56(b) displays the
relationship between duty cycle (%), and thus average power (W), and the absorption
properties of the surface. It can be observed that tuning the duty cycle of the pulse is
critical, as with lower duty cycle it was feasible to achieve the desired low absorption
characteristics. It is also clear from this figure that there is a proportional relationship
between the laser duty cycle and the absorption. Longer duty cycles result in more
energy used for the laser process, directly causing a more significant heat affected
zone (HAZ). This more intensely modified surface directly contributes to the increased
absorption of light which in turn would result in less light being available to the active
solar cell medium for photocurrent generation.
99
Figure 56. (a) Effect of Traverse Speed with PRF=400Hz and Duty Cycle=6%. (b)
Effect of Duty Cycle with PRF=400Hz and Traverse Speed=8.333mm/s.
5.2 Design of Experiments and Response Surface Methodology (RSM)
The R-squared value obtained for this statistical analysis was 0.994 and the model F-
value was 116.48, meaning that the model was significant and good correlations
between process parameters and process outputs existed. Further information on the
interpretation of these values is available in Appendix B. Figure 57 shows the
predicted vs. actual values for the response surface model created using the
experimental data. This graph shows that the generated model fit the experimental data
very well. If the data points are located along the straight line, it is a sign that the
predicted values are close to the actual experimental values.
Figure 58 displays the relationship between the two most influential factors in the laser
process; traverse speed (mm/s) and duty cycle (%). It can clearly be observed that
there is a non-linear relationship present over the investigated parameter range. The
parameter interaction in this case has resulted in lowest absorption at high traverse
speed and low duty cycle. The parasitic absorption loss under the AM1.5G spectrum
was measured to be between 2.27% and 63.28%. This was largely attributed to the
rough surface morphology remaining after laser processing.
100
Figure 57. Values predicted by the model equation against the actual
values. Points located along the line indicate the model provides an
acceptable fit.
Figure 58. RSM surface plot showing the correlation between the
significant input parameters and the resulting parasitic absorption in the
processed material (PRF=280Hz).
101
5.3 ANOVA Analysis
Model equation
The regression co-efficients for the quadratic model fitted to the experimental data are
presented in Table 26.
Table 26. The values of the Regression Co-Efficients for the quadratic
response surface model shown in Figure 58.
Co-efficient Value
b0 15.73712
b1 -0.06562
b2 -1.87792
b3 -0.88093
b4 -5.4E-05
b5 0.014494
b6 0.44665
b7 9.41E-05
b8 0.090434
b9 -0.12789
Table 27 presents the results of the Analysis of Variance and the calculation of the
contributions of the most significant parameters to the overall model. It can be seen
that x3 (Duty cycle) provides the largest contribution to the model, while x1 (PRF)
provides the least contribution of the 3 main parameters.
Table 27. Contribution of the most significant factors to the quadratic
response surface model.
Contribution of most significant factors:
Factor x1 x2 x3 x1 x3 x2 x3 (x3)2
% Contribution 0.5 9.8 43.91 7.0 13.6 17.2
p-Value 0.0631 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
102
The R-squared results for the model are presented in Table 28. The difference between
the Adj R-Squared and the Pred R-Squared is less than 0.2, the model is considered to
only have significant terms included. In this case the difference between the two values
is 0.0381, which would indicate that all the terms of the quadratic model are
significant.
Table 28: R-squared values for the RSM model
R-Squared 0.9692
Adj R-Squared 0.9528
Pred R-Squared 0.9147
Adeq Precision 27.719
5.4 SEM and EDXS Characterisation
The SEM images shown in Figure 59(a) and Figure 59(b) agree with the results
presented in the model. A significant surface modification can be seen Figure 59(b),
due to the low processing speed and high duty cycle. In contrast to this, the surface
presented in Figure 59(a) is significantly smoother. The smoother surface can be
attributed to the lower duty cycle and higher processing speed which reduced the
overall laser energy absorbed by the surface as the beam traverses it. As less laser
energy was absorbed, less surface material was ablated, which in turn results in less
recast on the surface. Less ablation also resulted in a shallower channel in comparison
to when a higher duty cycle was utilised.
Figure 60(a) and (b) show the cross-sections of the maximum and minimum texture
depths achieved by laser processing. Some cracking due to thermal shock is visible at
the bottom of the groove in Figure 60(a). The size of the textures varied from 72-974
μm in depth, and 146-213 μm in width at the widest point. Due to the ablation-
vaporisation mechanism utilised in creating the micro-grooves, the surface roughness
could be seen to increase with increasing laser power, as shown in Figure 59. This can
be attributed to increased material removal rates and the re-solidification of material
on the surface from cooling vapour.
103
Figure 59. Top down view of laser micro-machined channels in Fused Quartz. (a)
Sample 21 with low absorption properties (PRF=400 Hz, Speed=8.3 mm/s, Duty
Cycle=2%). (b) Sample 9 with high absorption properties (PRF=400 Hz, Speed=1.7
mm/s, Duty Cycle = 6%).
Figure 60. Cross sectional view of the groove texture.(a) Sample 16 with the
maximum (duty cycle 6%, PRF 160 Hz, speed 5 mm/s) and (b) Sample 21 with the
minimum texture depths achieved by laser ablation (duty cycle 2%, PRF 400 Hz,
speed 8.33 mm/s).
104
A summary of the channel dimensions found in this study are available in Appendix
F. The average channel depths, widths at top, and widths at 50% depth are presented
in Figure 61, Figure 62, and Figure 63 respectively. The dimensions of laser
micromachined channels in glass have been analysed extensively in other studies
using ANOVA analyses [44, 81, 125]. This type of analysis has not been performed
here as it is not a main objective of this work, however the results in this study correlate
well with those reported in those studies.
Figure 61: Average channel depths.
From the channel depth analysis it can be seen that the depth is quite repeatable at the
measurement locations along the channels. The largest standard deviation reported
across the three measurements is 31 μm, while the majority of the other standard
deviations are below 20 μm. This level of repeatability is a good indication that the
manufacturing process is repeatable and stable. The largest average channel depth
measured was 974 μm, while the shallowest was just 72 μm.
0
200
400
600
800
1000
1200
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Aver
age
Chan
nel
Dep
th (
µm
)
Sample / Channel
105
Figure 62: Average channel widths at the top surface.
From the channel width (top) analysis it can be seen that the width at the top of the
channels only varies within a small range, from 169 μm to 208 μm. This can be
attributed to the fact that the laser beam size at the focal spot is diffraction limited, and
is not dependant on the laser parameters investigated in this work. A diffraction limited
laser spot is the smallest size to which a laser beam can be focussed, where optical
diffraction effects are the limiting factor as opposed to other optical aberrations. As
the laser processing Average Power (W) is reduced (by reducing the Duty Cycle %),
the amount of laser irradiance (W/m2) delivered to the workpiece is also reduced. This
causes a reduction in the width of the channel at the surface, as there is less thermal
energy available to ablate the full width of the laser focal spot.
0
50
100
150
200
250
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Aver
age
Ch
ann
el W
idth
at
top
(µ
m)
Sample / Channel
106
Figure 63: Average channel width at 50% depth.
From the channel width (50%) analysis it can be seen that the width at the 50% channel
depth point is inversely proportional to the channel depth. This can be attributed to the
fact that the deeper channels have quite steep sidewalls, as is shown in Figure 60(a).
This channel profile can be attributed to a different thermal conduction mechanism
deeper in the channel than is found on the surface [81]. There may also be an effect
from the directionality of the laser assist gas flow, which causes the depth to be more
directional at higher laser powers. Nevertheless, this measurement at 50% channel
depth does give some insight into the overall aspect ratio of the channel.
Figure 64 presents a 3D display of channel geometry for sample 21 which was
fabricated using the lowest duty cycle and highest traverse speed. The depth of the
channel is 70-80 micrometres. In contrast, channels fabricated using the highest duty
cycle had depths approaching approximately 1 millimetre. This 3D profile was
recorded using the 3D non-contact optical profilometer developed during this work.
0
50
100
150
200
250
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Aver
age
Ch
ann
el W
idth
at
50
% D
epth
(µ
m)
Sample / Channel
107
Figure 64: 3D surface profile of sample 21, 70-80 micrometres in depth,
as captured by the 3D optical profilometer.
Figure 65 (a) and (b) are the corresponding EDXS spectra corresponding to the SEM
images in Figure 59(a) and (b) respectively. Over the 10 randomly selected points for
the EDXS analysis the average increase of silicon at the surface was 5% for Figure
59(a) over Figure 59(b). This change in composition could also be factor in the light
absorption characteristics of samples processed with higher duty cycles. It can clearly
be seen that the original untextured reference surface composition consisted of
significant amount of oxides on the surface (Figure 65(c)). The results of the EDXS
measurements are presented in Table 29.
108
Figure 65. EDXS results for (a) Sample 21 with low absorption, (b)
Sample 9 with high absorption, and (c) untextured Reference Sample
Table 29. Comparison of samples with highest and lowest light
absorption properties.
Analysis Sample 21 Sample 9 Reference
% Absorption 2.25% 61.82% 0%
Morphology Smooth shallow channel
with less recast
Rough dimpled surface
structure
‘As received’
untextured Surface
EDXS
(atomic %)
Si: 35.1%,
O: 64.9%
Si: 40.5%,
O: 59.5%
Si: 15.1%
O: 84.9%
109
5.5 Effect of HF etching on absorption of light
As both the surface morphology and surface composition of the laser processed
surface may influence the parasitic light absorption characteristics, an investigation
into the effect of post-processing the textured surfaces using a hydrofluoric acid etch
was completed. As presented in Figure 66, acid etching the laser treated samples
greatly reduces the absorption of light at the surface. Figure 66 presents the average
absorption of the light across the AM1.5G spectrum. All of the samples in this
investigation, with the exception of sample 21, were found to absorb less light after 5
minutes of etching in HF.
Figure 66. The effect of hydrofluoric acid (40%) etching on the
absorption of light at the surface.
An investigation into the effect of HF etching on the surface morphology of the laser
treated samples was also carried out. Electron microscopy and optical microscopy
were utilised to analyse the sub-micrometre surface features and how their structures
were altered with respect to the amount of time the acid etching was carried out. The
unetched laser processed surface of sample 21 is presented in Figure 67. It can be
observed that there are a range of surface features, which have been measured by SEM
to range from about 400 nm to several microns in size. It is these structures that were
110
of interest as they are of the order of wavelength of visible light and can contribute to
Mie Scattering within the superstrate [126]. The effect of HF etching on the surface
morphology is visible in the images presented in Figure 68(a) and (b). It was observed
that the acid etching dulls the appearance of these sub-micrometre structures, with the
longer etching times causing the features to effectively elongate.
Figure 67. An electron microscope image of the surface of sample 21
before acid etching. It is possible to see a large amount of re-cast
material. Visible feature sizes range from approximately 400 nm to
several microns.
Figure 68. (a) Surface of sample 21 after 5 minutes HF etching, and (b)
Surface of sample 21 after 10 minutes etching.
111
By acid etching with hydrofluoric (HF) acid for periods of 10 minutes it was possible
to further observe the smoothing effect of the rough surface morphology as shown in
Figure 69 [127, 128]. As the increase in parasitic absorption is attributed to increased
laser power and vapour deposition during processing, surface material removal by HF
treatment provided the highest benefits to the textures created using the higher powers.
While most samples displayed improved transmission characteristics after etching, for
the two samples (samples 20 and 21) with the lowest parasitic absorption losses, the
effects of acid etching on transmission were negligible (<1%) as shown in Figure 66.
a)
b)
c)
Figure 69. Top down view of the surface morphology of laser ablated
channels (duty cycle 4%, PRF 400Hz, speed 1.67mm/s). unetched (a),
after etching for 10 minutes (b) and after etching for 20 minutes (c).
112
Figure 70 illustrates the change in light absorption with respect to acid etching time.
It can be observed from this data that there is a shift in the absorption peak towards
the red end of the spectrum with increasing etching time. The EDXS results presented
in Figure 71 and Table 30 indicate that there was little change in the surface chemistry
of the laser textured glass after HF etching. This indicates that the increase in the
absorption of light at the surface after laser processing may be attributed more
significantly to the elongation of the acid etched features.
Figure 70. Absorption of AM1.5G spectrum by sample 21 with respect to
etching time (data fitted with 6th order polynomial trend line for clarity).
It can clearly be seen that etching causes a shift in absorption towards the
red end of the spectrum. This can be directly correlated to the elongation
of nanostructures on the surface due to acid etching.
Figure 71. EDXS spectrum of sample 21 after etching with HF.
Composition is close to that before HF etching (Figure 65(a))
113
Table 30 Comparison of surface composition before and after HF acid
etching.
Analysis Sample 21 before HF etching Sample 21 after HF etching
% Absorption 2.25% 3.68%
Morphology Smoother surface with
elongated nanostructures
Smooth shallow channel with
less recast
EDXS
(atomic %)
Si: 33.2%,
O: 66.8%
Si: 35.1%,
O: 64.9%
5.6 Efficiency vs angle of incidence
In order to assess the possible gains of the textured superstrate it was necessary to test
the efficiency of a solar cell used with the textured superstrate with varying angular
incidence of light. Figure 72 displays the acquired I-V data for the same solar cell
mounted under the three different superstrates; one untextured as a reference and the
other two with the maximum transmission textures (samples 20 and 21). This data was
acquired with the light incident at 10° to the surface. The texture characteristics and I-
V data are presented in Table 31. The results indicate that at acute angles of incidence
there is a significant improvement in the cell performance due to the enhanced light
trapping effect caused by the textures on the superstrate. At 10° incidence to the
surface plane a relative efficiency gain in the solar cell of 51% was found for sample
21 in comparison to the untextured reference. Voc, Isc, and fill factor were also
improved relatively by 2.1%, 44.6%, and 2.7% respectively.
Figure 72. I-V curves for solar cell at the acute angle of 10° incidence.
untextured reference (solid), sample 20 (dotted), and sample 21 (dashed).
0
0.0005
0.001
0.0015
0.002
0.0025
0 0.1 0.2 0.3 0.4 0.5 0.6
Cu
rre
nt
(A)
Voltage (V)
114
Table 31. Parameters and results for I-V measurement with light incident
at 10° to surface.
Sample Untextured Sample 20 Sample 21
Depth (μm) - 95 72
Width (μm) - 173 146
VOC (V) 0.478 0.483 0.488
ISC (A) 0.00139 0.00185 0.00201
Fill Factor (%) 67.20 67.15 69.01
Efficiency (%) 0.49 0.66 0.74
While there were significant efficiency gains at an angle of incidence of 10°, the solar
cell efficiency was degraded relative to the untextured reference at higher angles of
incidence. Figure 73 indicates that cell efficiency was only improved relative to the
untextured reference for acute angles in the range of 20° to 10°. As shown in Figure
60(b) the surface of the samples tested have a somewhat sinusoidal shape. According
to simulations performed by Sanches-Illescas et al. the highest benefit of the textured
surface can be found at acute angles of incidence, which corresponds well with our
findings [51]. The reduced performance at other angles could be attributed to the effect
of laser processing on the surface quality itself, changes in surface composition, and
also the increased path-lengths travelled by the light due to enhanced scattering,
causing additional parasitic absorption losses in the superstrate.
Figure 73. Solar cell efficiency with different angles of incidence for the
untextured reference (solid), sample 20 (dotted), and sample 21 (dashed).
0
1
2
3
4
5
6
7
8
90 80 70 60 50 40 30 20 10
Effi
cie
ncy
(%
)
Angle of Incidence (°)
115
Chapter 6 Characterisation for
Quantum Dot Solar Cells
Quantum dot based solar cells are an emerging field of research. Quantum dots offer
great advantages in that they can be low-cost to produce, yet be tailored to optimise
conversion efficiencies through size control. For the first part of the investigation of
such a solar cell set-up, the optical absorption characteristics of alkyl-capped silicon
quantum dots were investigated. It is envisaged that these quantum dots could be
utilised in next generation solar cells. The second section of this investigation
concentrates on the 3D characterisation of macroporous monolithic morphologies,
which have wide applications in existing and next generation solar cell technologies.
This is especially true for quantum dot based solar cells, where much work has
concentrated around the use of porous semi-conductors on glass superstrates.
6.1 Absorption characteristics of alkyl-capped silicon quantum dots
Figure 74 shows the optical microscope image of the silicon nanocrystals taken
immediately after a third repetition of the drop deposition technique. The silicon
quantum dots appear to be about 1 µm in diameter, possibly due to diffraction
broadening, significantly larger than shown by other topographical characterisation
methods, such as AFM as shown in section 3.16.2. This could be attributed to the
formation of clusters of nanocrystals on the surface of the substrate after subsequent
depositions [117]. It is clearly visible from this image that the Si nanocrystals do not
form a continuous layer on the substrate material. It can also be observed that there
are varying degrees of surface coverage of the Si-QD with some areas more and some
less sparsely populated. Surface coverage could likely be improved through
optimisation of the Si-QD solution concentrations and the deposition method.
116
Silicon Quantum Dot clusters
after deposition.
Bubbles in
evaporating solvent
50 µm
Figure 74: Optical microscope image of nanocrystalline silicon on
substrate shortly after drop deposition. Bubbles indicate that the solvent
is still present.
Figure 75 shows the absorption characteristics of the silicon quantum dots after they
were bonded to the glass substrate. The absorption measurement was performed after
each repetition of the drop deposition technique and subsequent drying time. It can
clearly be seen that there is a small increase in absorption of the incident broadband
light by the deposited nanocrystals between 420 nm and 520 nm. Increases in
absorption in the same wavelength range have been found in studies where Si
nanocrystals are embedded in a SiO2 matrix [64, 65]. Other studies on alkyl-capped Si
nanocrystals have found an increase in absorption only from approximately 325 nm
and below [63, 129]. Vasiliev et al. simulated the effect of oxides on the optical
absorption of Si nanocrystals using a confinement model revealing that for certain
oxide configurations, the optical absorption gap was further lowered to 2.4 eV (520
nm), which correlates well with our results [130]. This indicated that it is the presence
of oxides in our nanocrystals, as indicated by the XPS analysis performed by
collaborators [115], was causing absorption at lower photon energies. However, it was
not possible to achieve a uniform coverage of the substrate with the Si quantum dots,
which may have led to further inconsistencies in the absorption measurement in this
preliminary investigation. The source of this error is due to inhomogeneous
distribution of the quantum dots, and is also due to the variation in the illumination
117
source intensity in the integrating sphere setup. The low population of silicon quantum
dots on the substrate surface led to a lower signal to noise ratio during the
measurements. By increasing the number of quantum dots on the substrate between
each measurement an improved signal would be obtained. The variation in light source
intensity could be corrected by creating several samples with different quantum dot
populations and performing all of the absorption measurement in one session.
Figure 75: Light absorption characteristics of Si QDs after 1 deposition
(green), 2 depositions (blue), 3 depositions (red). The absorption of the
glass and the solvent have been subtracted from this data.
6.2 Absorption characteristics of alkyl-capped silicon quantum dots on laser
textured superstrates
Taking the absorption results of the previous section, it was of interest to investigate
the absorption characteristics of the silicon quantum dots when used in combination
with laser textured superstrates, as investigated in Chapter 5. Three samples which
exhibited low parasitic absorption losses after HF etching (Samples19, 20, and 21)
where chosen for this investigation. The absorption results for the three different
textured superstrates are shown in Figure 76.
0
5
10
15
20
420 520 620 720 820 920
Ab
sorp
tio
n
%
Wavelength (nm)
118
Figure 76: Light absorption characteristics of Si QDs on laser textured
superstrates after drop casting. Untextured reference (blue), Sample 19
(red), Sample 20 (green), and Sample 21 (violet).The absorption of the
glass and the solvent have been subtracted from this data.
The reference measurement itself indicated increased absorption between 420nm and
520nm when compared with the results from the initial absorption study, which can
be attributed to the fact that five depositions of silicon quantum were performed for
this measurement in an effort to increase the sensitivity and resolution of the results.
Compared with the untextured reference superstrate, all three textured superstrates
indicated increased absorption across the entire spectrum from 420nm to 1000nm.
However, as was found in the initial absorption characterisation experiments, the
silicon quantum dots used in this investigation absorb light mostly in the 420 nm to
520 nm range. The absorption across the remained of the spectrum from 520 nm to
1000 nm in this case could be attributed to several possible causes. Firstly, during
deposition time was allowed for the solvent to evaporate. However, due to the
increased amount of silicon quantum dots depositions (five depositions for this
experiment), it is possible that, despite best efforts, some solvent remained on the
superstrate at the time of measurement as there remained a ‘waxy’ residue on the
surface. Secondly, the effect of light scattering caused by the laser textured superstrate
0
10
20
30
40
50
60
420 520 620 720 820 920
% A
bso
rpti
on
Wavelength (nm)
119
may have caused additional absorption in the remaining solvent. In order to
specifically investigate the increase in absorption between 420 nm and 520 nm, the
measurements were referenced relative to each other to each other between 520 nm
and 1000 nm, which contributed to the negation of the unaccounted effects of
absorption in any unevaporated solvent. The highest increase in absorption over the
420 nm to 520 nm range was found for Sample 19, the spectrum of which is indicated
in Figure 77. An approximate increase of 5% in absorption of across this wavelength
region was found. While the gain is marginal, there is an indication that the laser
textured glass does contribute to enhanced light absorption in the deposition silicon
quantum dots. An increase in absorption in this wavelength region correlates well with
the initial absorption spectra measured, and hence it could be deduced that the increase
in absorption measured is due to increased absorption in the quantum dots and not just
in any remaining solvent.
Figure 77: Spectra for untextured reference (blue) and Sample 19 (red).
The spectra have been reference matched from 520 nm to 1000 nm.
6.3 FIB serial sectioning for the 3D reconstruction of macroporous structures
Significant amounts of work have been done in the research of porous silicon for
photovoltaic applications, as there are opportunities for enhanced light trapping
performance when utilising the porous morphology architecture. Next generation solar
cells, such as dye-sensitised solar cells and certain quantum dot solar cell architectures
0
5
10
15
20
25
30
420 520 620 720 820 920
% A
bso
rpti
on
Wavelength (nm)
120
also make use of macro- and nano-porous structures. These structures can play various
roles depending on the exact cell type and design, however, one commonly cited effect
of utilising such structures are the beneficial light trapping effects they provide. As
part of this study, a method for characterising macro-scale porous structures in 3D was
investigated utilising a focused-ion-beam (FIB) to serially section the sample, while
simultaneously being able to image the sample cross section at high magnification
using scanning electron microscopy (SEM). The materials investigated in this study
were carbon and silica based monoliths. While these materials would potentially have
some applications in solar cell architectures, the aim of work was to investigate the
usefulness of the characterisation method for these structures, as opposed to
specifically for these materials.
6.4 3D reconstruction and morphological analysis of the carbon monolith
The resulting 3D reconstruction of the porous carbon monolith from the 2D-image
stack consisting of 93 images is shown in Figure 78. The volume reconstructed had a
rectangular geometry with dimensions of 58.2 μm x 26.5 μm x 9.3 μm in the X, Y and
Z directions, respectively. As shown in Figure 78, the reconstructed carbon monolith
exhibited an interconnected porous structure formed by quasi-spherical macropores
with sizes mostly ranging from approximately 4 μm to 7.5 μm. Moreover, the 3D
reconstruction revealed a non-homogeneous spatial distribution of the inter-skeleton
macropores within the volume analysed. The latter could be attributed to a non-
uniform packing density of the silica particles within the phenolic resin used for
preparation of the carbon monolith. Likewise, the relatively broad range of macropore
sizes observed in the monolith reconstruction could be attributed to the use of silica
particles with non-uniform sizes as templates, which could also largely affect the
packing density of the silica particles in the silica/phenolic resin mixture.
121
Figure 78: 3D reconstruction of the porous carbon monolith
Furthermore, the presence of pores with sizes between ca.10 μm and 15 μm range is a
clear indication of the formation of voids in the space between the silica particles
during preparation of the monolith. This could be explained by the limited infiltration
of the phenolic resin into the inter-particle space due to poor mixing of the silica
particles with the phenolic resin and/or high viscosity of the phenolic resin during
preparation. This non-homogeneous macro-pore morphology is consistent with
previous results obtained by Eltmimi et al. in carbon monoliths prepared with 5μm
silica particles as templates, where the presence of inter-particle voids was already
suspected based on experimental evidence [123].
Therefore, the direct confirmation of the real morphology of a monolith via
reconstruction and visualisation of its porous structure (including voids) represents a
clear advantage of this FIB-SEM methodology, which can significantly contribute to
speeding up the optimisation of the porous structure morphology in view of improving
its applications specific performance. The total macroporosity of the volume analysed
following binarisation of the corresponding image stack was found to be 72.9%. This
value agrees well with the results obtained on the same sample type by MIP, where
the total porosity was found to be 76.8% and 80.2% for two consecutive analyses. MIP
measurements also allowed determination of the pore size distribution, as shown in
122
Figure 79. A median pore diameter (in volume terms) of 2.6 μm and 2.9 μm was
obtained for the two consecutive MIP analyses, respectively. This diameter range was
well within the size range suitable for manual outlining when processing the images,
which is why the porosity values correlate well.
Figure 79: Mercury porosimetry results showing pore size distribution in
the carbon monolith [122].
6.5 3D reconstruction and morphological analysis of the silica-based monolith
Figure 80: 3D reconstruction of the porous silica monolith (Monotrap).
123
The 3D reconstruction of a commercial porous silica-based monolith from the
corresponding 150-image stack obtained by FIB-SEM analysis is shown in Figure 80.
The reconstructed volume also had a rectangular geometry with dimensions of 44.7
μm x 33.7 μm x 10 μm in the X, Y and Z directions, respectively. As can be seen in
Figure 80, the reconstructed volume exhibited a rather homogeneous spatial
distribution of interconnected macropores with irregular shapes and variable sizes,
which contrasts well with the mainly spherical shaped pores found during the analysis
of the carbon monolith as shown in Figure 78. Therefore, the modification of the silica
monolith skeleton with activated carbon proved to be highly effective for successful
reconstruction and visualisation of the silica-based monolith macropore structure by
FIB-SEM analysis. The electrical conductivity provided by the carbon assisted in the
minimisation charge build-up during SEM analysis, typically arising in non-
conducting materials such as silica, and successfully allowed FIB-SEM analysis.
Although no specific information on how this commercial carbon-modified silica-
based monolith was produced, protocols for the preparation of carbon-silica
composites have been already described in previous works. The total macroporosity
of the volume analysed for the carbon-modified silica-based monolith was found to be
52.9%, which agrees well with the 56.0% value for total porosity obtained by MIP.
The pore size distribution of the silica-based monolith can be seen in Figure 81, with
a median pore diameter (in volume terms) of 2.7 mm.
Figure 81: Mercury porosimetry results showing pore size distribution in
the silica-based monolith [122].
124
Overall, the results of this investigation indicate that FIB-SEM serial sectioning is a
useful method by which to evaluate the 3D morphology of porous materials. This
study has found that the FIB-SEM method has comparable accuracy in terms of
porosity evaluation of porous materials where the pore size distribution it above 1 μm
diameter. One of the main advantages of this characterisation method is that it can
evaluate the true porosity of a material, which includes both open and closed pores,
whereas other characterisation methods such as MIP do not offer this functionality.
This especially has application in the evaluation of porous materials in solar cells, as
in the case with materials such as porous TiO2, closed pores may have a great effect
on the light scattering and electron transport mechanisms within the solar cell.
While the FIB-SEM approach definitely has merit in the evaluation of the porous
materials, there also remain some challenges with this method. Firstly, there is a high
cost associated with the process as it involves high vacuum, and it is also time
consuming. Secondly, there remains some research to be completed on the use of the
FIB-SEM approach with higher SEM magnifications. This is especially true for non-
conducting materials such as silica, where the sample can become charge saturated at
high magnifications, which reduces SEM image quality considerably. Another
challenge lies in the fact that at high magnification, the depth-of-field of the SEM is
reduced which makes it even more time consuming.
125
Chapter 7 Conclusions
Verification of the 3D non-contact optical profilometer
Verification investigations performed on the developed 3D noncontact optical
profilometer included measurement of the linear stage performance using a dual scale
analysis, and the evaluation of the 3D measurement performance. The analysis of the
linear stage performance was performed in line with ISO 230-2 standards for the
calculation of stage accuracy and repeatability. The investigation confirmed the
suitability of the design for 3D profilometry use. The performance of the stepper motor
axes designed was considered to be acceptable considering the low-cost nature of the
design. The stepper motor driven X and Y axes performed considerably better in the
small scale analysis, with bi-directional accuracies of ±8 μm and ±15 μm, which can
be attributed to the shorter travel range of the investigation and also the superior
accuracy of the dial gauge indicator utilised in the small scale measurement analysis.
There XY positioning accuracies are acceptable over this measurement range for the
purposes of image stitching if sufficient image overlap is allowed as to enable software
to perform accurate alignment of the measurement data. The Z-axis performed with
acceptable uni-directional accuracy of 10 μm over the 200 μm small scale analysis
displacement. When considering the possible sources of measurement error in this
exercise, it would be worthwhile suggesting that an alternative method of stage
measurement, such as laser interferometry, be implemented to gain a better
understanding of the various aspects of the linear stage performance. Laser
interferometry would provide a method of measurement over the full travel range with
nanometre level accuracy, data which can then be used to calibrate the system to a
much higher level of accuracy.
To evaluate the 3D measurement performance of the system, a procedure was
implemented to evaluate the performance over a dual scale analysis. As it was difficult
to source a suitable traceable calibrated step height reference in this size range and
with the appropriate surface characteristics for this type of optical instrument it was
necessary to design and manufacture a purpose build step height artefact. This was
achieved by milling a brass measurement artefact to the desired dimensions, with a
126
nominal 100 μm step height. This sample was subsequently measured by Scanning
Electron Microscope (SEM), White Light Interferometry (WLI), and Stylus
Profilometry (SP). These three measurement methods returned almost identical step-
height measurement results of 103 μm. A nominal 500 μm thick metrology gauge
block was utilised as the other sample in the dual scale analysis, and was also measured
using WLI and SP, with the resulting step height of 512 μm. While these
measurements are not traceable metrology standards for the evaluation of the absolute
accuracy of the system, they were designed to give some initial indication of the
estimated accuracy of the system. The investigation demonstrated the capability of the
developed system to measure step heights in the range of 100 µm to 500 µm, however
the evaluation of the absolute accuracy of the system would require significantly more
work around the manufacture and statistical analysis of the measurement artefact. The
accuracy evaluation of depth-from-focus based profiling instruments is on-going work
within the field of metrology, where National Metrology Institutes have suggested that
further investigation and standardisation is required [111]. These artefact
measurement values were then used in the evaluation of the 3D measurement
performance of the developed profilometer. Results were calculated from a Design of
Experiments carried out using different Z-axis sampling intervals, with step height
reported after data post-processing. Accuracies reported on the measurement of the
103 μm step-height were as good as 99.95%, which gives initial indication that the
performance of the system is within an expected range. While acknowledging that this
accuracy measurement is only indicative and not absolute, one can equate this
performance to a measurement uncertainty of ±4 µm over the 103µm step height. This
value correlates well with the with the 10 μm uni-directional accuracy performance of
the the Z-axis in the initial linear axis performance investigation which was carried
out over a 200 μm travel range. These performance characteristics compare well with
those reported by Sun and Claverley who performed calibration analyses on a
commercially available instrument with a similar principle of operations, whose
measurement of spherical reference samples resulted in 3D measurement accuracies
in a range between -7 μm and +19 μm [111]. The methods of step height evaluation
reported in this work contribute to the analyses of depth-from-focus based
measurement instrumentation and to the calibration and verification issues reported by
Sun and Claverley.
127
Measurement repeatability was another output of this investigation. When the absolute
accuracy of an artefact cannot be certified, the same artefact can be used to assess the
repeatability of the measurement technique through the repetition of measurements.
This investigation found the largest standard deviation of the mean plane heights
across six measurement runs to be between 0.1 µm and 0.9 µm. This compares well
with Sun and Claverley’s investigation, which found the standard deviation of sphere
size measurements to be in the sub-micrometre range [111]. It can be derived from
these repeatability measurements that the developed device may be suitable for the
evaluation of structures with height in the sub-10 µm range, however further
investigation is required in order verify this and to establish what measurement
conditions are most favourable for this outcome.
It has been reported that 3D profilometry instruments based on the depth-from-focus
principle have limitations in the measurement of surface topographies of translucent
materials, particularly in the presence of micro-scale surface roughness [80].
However, in this work it has been clearly demonstrated that the 3D non-contact optical
profilometer developed in this thesis offer capabilities in the measurement of laser
micro-machined translucent materials in the presence of micro-scale surface
roughness, which is a new result in this research field.
It is envisaged that this device can be utilised in gaining further understanding of
surface micro-structures. The usefulness of this device includes the surface
topography analysis of various micro-structures used in solar cell applications. Using
the 3D non-contact optical profilometer to measure the surface topography of laser
structured glass for example, would allow for real surface topologies, including
roughness, to be characterised and used in simulations of the light trapping effect in
solar cells. This approach should therefore provide a new level of highly accurate data
and simulation realism to this research field.
128
Laser texturisation for enhanced light trapping performance
Laser textures were created on fused silica glass using a CO2 laser operating in pulsed
mode, for which the parameters were defined by Design of Experiments. The size of
the textures varied from 72 μm to 974 μm in depth, and 146 μm to 213 μm in width.
The parasitic absorption loss of light in the laser textured glass was measured using an
integrating sphere and spectrometer setup, and losses ranged from 2.25% to 63.28%.
The surface morphology was smoothed by HF etching for up to 20 minutes and
resulted in improved transmission for most samples. The increase in transmission after
acid etching of the two samples with the lowest parasitic absorption losses (samples
20 and 21) were negligible. The I-V data for the solar cell under different textures was
assessed by encapsulating the solar cell in transparent resin and placing it in contact
with the flat side of the laser textured front cover glass. This allowed the measurement
of the characteristics of the solar cell under the textured superstrate by measuring the
I-V data at varying angles of incidence. While increased performance was not found
at all angles of incidence, for acute angles of incidence (10°) the relative efficiency
increased by up to 51% with an increase in the Voc of 2.1% and in the Isc of 44.6%. An
improvement in the fill factor of 2.7% was also observed.
While these results indicate that this particular laser processing approach is not
favourable at all angles of incidence as experienced in an everyday practical situation,
the evidence does point to the fact that there have been some improvements achieved
with this process. It is postulated that by using different laser sources more suited to
the fine ablation control required to further optimise this process, that much improved
solar cell performances could be achieved. Indeed, when considering the results
reported by Kang et al., it is certainly feasible with a different laser source, and with
improved glass processing, that building integrated solar photovoltaics would benefit
from micro-structured solar cell superstrates [38].
129
Characterisation for Quantum Dot Solar Cells
From the perspective of absorption in next generation solar cell materials, the
absorption of light by alkyl-capped silicon quantum dots was investigated. The
absorption of the alkyl-capped Si quantum dots drop deposited onto a 1 mm thick
fused quartz substrate was measured using an integrating sphere and spectrometer
setup. Subsequent depositions indicated that there is an increase in absorption of
broadband light in the 420 nm to 520 nm range. However, the non-uniformity of the
drop deposited layer may have led to inconsistencies in the absorption measurement.
Much work needs to be done in order to control the layer by layer depositions and to
understand the properties of this novel material. By using other deposition techniques,
such as evaporation, it will be possible to control the thickness of silicon quantum on
the surface of the substrate, which may improve the understanding of the material
properties.
Subsequent investigations were completed to evaluate the effect of light trapping when
the silicon quantum dots are deposited directly on a laser textured superstrate.
Deposition cycles consisting 4 ml drops on to the target for five cycles were prepared
and evaluated. Additional absorption across the spectrum from 420 nm to 1000nm was
observed, however the increase in absorption from 520 nm to 1000 nm is thought to
have originated from increase light trapping within solvent remaining on the
superstrate. When the data was adjusted to allow for this effect, an increase in
absorption of up to 5% in the 420 nm to 520 nm was observed. This can be attributed
to additional light trapping effects in the silicon quantum dots and fused quartz
superstrate. This increase correlated well with initial observations on the absorption
characteristics of these silicon quantum dots, and represents novel results in the field.
As porous material morphologies are increasingly common in the design of next
generations solar cells, an investigation into a 3D characterisation method which could
be used to gain a better understanding of the pore connectivity and topology. To
achieve this a Focused-Ion Beam (FIB) milling technique was utilised to provide serial
sectioning capabilities within a Scanning Electron Microscope (SEM). Both
conducting and non-conducting materials were investigated, in the forms of a carbon
monolith and silica monolith respectively. The serial sectioned images were post-
130
processed utilising software developed to correct for dimensional anomalies due to the
imaging angle of the SEM. After manual image segmentation and subsequent 3D
reconstruction, the porosity of the materials was analysed, and the results compared
directly with those of an established porosity measurement technique, Mercury
Intrusion Porosimetry (MIP). Porosity results for the carbon monolith were found to
be 72.9% and (averaged) 78.5% for the FIB-SEM analysis and MIP analysis
respectively. For the silica based monolith, porosity results were found to be 52.9%
and 56% for FIB-SEM and MIP analyses respectively. This indicates that the FIB-
SEM technique is a suitable technique for the evaluation of macroporous materials.
One main advantage of the FIB-SEM method is that it is possible to evaluate all of the
pores in a macroporous material, whereas traditional measurement techniques such as
MIP can only measure open and connected pore structures. In a comprehensive review
by Cocco et al. on three-dimensional micro-structure imaging methods for energy
materials, there was no report of the FIB-SEM characterisation method being utilised
for the characterisation of photovoltaic materials [131]. It is envisaged that this
characterisation method can be used to gain a better understanding of macroporous
solar cell materials, and also can be used as a real 3D surface measurement instrument,
data from which can be utilised in simulations of the interaction of solar cell materials
with light and the light trapping effect created by such porosity.
7.1 Future Work
There are several possible end-uses for the technology developed in this research.
While the main focus of this research lies in the improvement of light trapping for
photovoltaic cells for solar energy harvesting, the technology may also be transferable
to other fields. The geometrical light trapping effect is of great importance in thin-film
and quantum dot based solar cells also, mainly due to the fact that there is less light
absorbing material present. In order to preserve the potential efficiency of these types
of solar cells, it is necessary to improve the chances of absorption. In thin-film solar
cells, this takes place not only through the light trapping effect, but also through the
longer path lengths the photons must take to traverse the cell. Simulations in the
literature have shown that the light trapping effect can increase the effective thickness
of thin-film solar cell materials by up to 40 times [8]. There are other possible uses in
the field of optics, for example, improving the absorption efficiency across a wider
131
range of angles in optical detectors is also of interest. It is believed that by varying the
texture it will be possible to control the sensitivity of the optical detectors at certain
angles, which operate in a similar fashion to large scale photovoltaic devices.
The 3D non-contact optical profilometer developed in this work may be leveraged in
several approaches. This tool can be used to evaluate the surface morphology of laser
processed surfaces, to evaluate micro-sized features in 3D and to evaluate surface
roughness. This surface data can also be exported for simulation purposes, which will
allow for more realistic manufactured surfaces to be utilised in simulation
investigations. This is of particular importance where this simulation involves
interaction with light, as features and roughness on the microscale greatly influence
the characteristics of the interactions of that material. However, there remains further
work to improve the characterisation of the capabilities of the developed system, both
in terms of accuracy of the motion axes and accuracy of the overall 3D measurement
system. There are multiple possible sources of error which should be minimised, and
a more accurate linear measurement technique such as laser interferometry should be
utilised to improve the performance of the technique and to widen the understanding
of the influence of various sources of inaccuracy on the overall 3D measurement
performance of the system. The work in this thesis has provided initial results to verify
the feasibility of the technique, however further work must be undertaken to validate
the use of such an instrument and configuration as a metrology tool suitable for
measurement at the micrometre scale.
The evaluation of silicon quantum dots has shown great promise in this work. The
absorption spectra measured correlated well with the estimated material band-gap.
While the results found were positive initial results, further work is needed on the
deposition of the quantum dots on the superstrates, which will greatly improve the
understanding of the interaction of this novel material with light. Further work could
include the development of a quantum dot solar cell based on the alkyl-capped silicon
quantum dots used in this work, however, much work needs to be done in designing
and optimising a solar cell architecture.
132
In a similar way the method developed through the use of FIB-SEM serial sectioning
can be utilised for the evaluation of porous macro-structures. While more work must
be done on the development of this method to work with smaller pore diameters and
with larger volumes, the indications from the results in this work are that this method
is a suitable method for the evaluation of pores with diameters on the order of 1 μm
equivalent diameter and greater, with accurate results available in the measurement of
porosity. Therefore, it is suggested that future work leverage the developed FIB-SEM
work to improve the simulation realism through the utilisation of improved surface
models based on real morphological data.
133
References
[1] A. Luque and S. Hegedus. Handbook of Photovoltaic Science and Engineering,
2nd Ed 2011Available: http://prism.talis.com/dcu/items/595709.
[2] Fraunhofer Institute for Solar Energy Systems ISE, "Photovoltaics report,"
Fraunhofer, Freiburg., Aug 2015. 2015.
[3] S. J. C. Irvine. Materials Challenges: Inorganic Photovoltaic Solar Energy 2014.
DOI: 10.1039/9781849733465.
[4] Y. Hamakawa. Thin-film solar cells: Next generation photovoltaics and its
applications. 2004.
[5] A. J. McEvoy, L. Castaner and T. Markvart. Solar Cells: Materials, Manufacture
and Operation (2nd ed. ed.) 2012.
[6] K. A. Jackson. Silicon Devices: Structures and Processing 1998.
[7] J. Wu and Z. M. Wang. Quantum Dot Solar Cells (1st ed.) 2013.
[8] Jianhua Zhao, Aihua Wang, P. Campbell and M. A. Green. A 19.8% efficient
honeycomb multicrystalline silicon solar cell with improved light trapping. Electron
Devices, IEEE Transactions On 46(10), pp. 1978-1983. 1999.
[9] C. S. Solanki. Solar Photovoltaics: Fundamentals, Technologies and
Applications 2009Available: http://prism.talis.com/dcu/items/595702.
[10] A. Manickavasagan and H. Jayasuriya. Imaging with Electromagnetic
Spectrum: Applications in Food and Agriculture 2014.
[11] P. Würfel Dr. Physics of Solar Cells: From Basic Principles to Advanced
Concepts (2nd updated and expanded ed. ed.) 2009.
[12] American Society for Testing and Materials G173-03, "Standard Tables for
Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37°
Tilted Surface." 2012.
[13] International Standards Organisation ISO 9845-1, "Solar energy -- Reference
solar spectral irradiance at the ground at different receiving conditions -- Part 1:
Direct normal and hemispherical solar irradiance for air mass 1,5." 1992.
[14] National Renewable Energy Laboratory. (). Solar Spectral Irradiance: Air Mass
1.5. Available: http://rredc.nrel.gov/solar/spectra/am1.5/.
134
[15] G. Kumaravelu, M. M. Alkaisi, A. Bittar, D. Macdonald and J. Zhao. Damage
studies in dry etched textured silicon surfaces. Current Applied Physics 4(2-4), pp.
108-110. 2004. . DOI: DOI: 10.1016/j.cap.2003.10.008.
[16] L. A. Dobrzański, A. Drygała, K. Gołombek, P. Panek, E. Bielańska and P.
Zięba. Laser surface treatment of multicrystalline silicon for enhancing optical
properties. Journal of Materials Processing Technology 201(1-3), pp. 291-296.
2008. . DOI: DOI: 10.1016/j.jmatprotec.2007.11.278.
[17] H. B. T. Li, R. H. Franken, J. K. Rath and R. E. I. Schropp. Structural defects
caused by a rough substrate and their influence on the performance of hydrogenated
nano-crystalline silicon n–i–p solar cells. Solar Energy Materials and Solar Cells
932008. . DOI: DOI: 10.1016/j.solmat.2008.11.013.
[18] D. Iencinella, E. Centurioni, R. Rizzoli and F. Zignani. An optimized texturing
process for silicon solar cell substrates using TMAH. Solar Energy Materials and
Solar Cells, 87(1-4), pp. 725-732. 2005.
[19] P. Papet, O. Nichiporuk, A. Kaminski, Y. Rozier, J. Kraiem, J. -. Lelievre, A.
Chaumartin, A. Fave and M. Lemiti. Pyramidal texturing of silicon solar cell with
TMAH chemical anisotropic etching. Solar Energy Materials and Solar Cells,
90(15), pp. 2319-2328. 2006.
[20] E. Forniés, C. Zaldo and J. M. Albella. Control of random texture of
monocrystalline silicon cells by angle-resolved optical reflectance. Solar Energy
Materials and Solar Cells 87(1-4), pp. 583-593. 2005. . DOI: DOI:
10.1016/j.solmat.2004.07.040.
[21] G. Kumaravelu, M. M. Alkaisi and A. Bittar. Surface texturing for silicon solar
cells using reactive ion etching technique. Photovoltaic Specialists Conference,
2002. Conference Record of the Twenty-Ninth IEEE pp. 258-261. 2002.
[22] T. Sakoda, K. Matsukuma, Y. Sung, K. Otsubo, M. Tahara and Y. Nakashima.
Additional plasma surface texturing for single-crystalline silicon solar cells using
dielectric barrier discharge. Japanese Journal of Applied Physics, Part 1: Regular
Papers and Short Notes and Review Papers 44(4), pp. 1730-1731. 2005. Available:
http://dx.doi.org/10.1143/JJAP.44.1730.
[23] Z. Xu, Y. Yao, E. P. Brueckner, L. Li, J. Jiang, R. G. Nuzzo and G. Logan Liu.
Black silicon solar thin-film microcells integrating top nanocone structures for
broadband and omnidirectional light-trapping. -Nanotechnology 25(30), pp. 305301.
2014. . DOI: 10.1088/0957-4484/25/30/305301.
[24] H. L. Chen, K. T. Huang, C. H. Lin, W. Y. Wang and W. Fan. Fabrication of
sub-wavelength antireflective structures in solar cells by utilizing modified
illumination and defocus techniques in optical lithography. Microelectronic
Engineering 84(5-8), pp. 750-754. 2007. . DOI: DOI: 10.1016/j.mee.2007.01.026.
135
[25] Y. Ein-Eli, N. Gordon and D. Starosvetsky. Reduced light reflection of textured
multicrystalline silicon via NPD for solar cells applications. Solar Energy Mater.
Solar Cells 90(12), pp. 1764-1772. 2006. . DOI: DOI: 10.1016/j.solmat.2005.10.024.
[26] Christian Schmiga, Jan Schmidt, Axel Metz, Arthur Endrös, Rudolf Hezel,
"17.6% efficient tricrystalline silicon solar cells with spatially uniform texture,"
Progress in Photovoltaics: Research and Applications, vol. 11, pp. 33, 2003.
[27] S. K. Srivastava, P. Singh, M. Yameen, P. Prathap, C. M. S. Rauthan, Vandana
and P. K. Singh. Antireflective ultra-fast nanoscale texturing for efficient multi-
crystalline silicon solar cells. Solar Energy 115pp. 656-666. 2015. . DOI:
http://dx.doi.org.dcu.idm.oclc.org/10.1016/j.solener.2015.03.010.
[28] B. Parida, J. Choi, S. Palei and K. Kim. Nanopyramid formation by ag metal-
assisted chemical etching for nanotextured si solar cells. Transactions on Electrical
and Electronic Materials [[Online]]. 16(4), pp. 206-211. 2015. . DOI:
10.4313/TEEM.2015.16.4.206.
[29] E. Manea, E. Budianu, M. Purica, I. Cernica and F. Babarada. Technological
process for a new silicon solar cell structure with honeycomb textured front surface.
Solar Energy Materials and Solar Cells 90(15), pp. 2312-2318. 2006. . DOI: DOI:
10.1016/j.solmat.2006.03.036.
[30] E. Manea, E. Budianu, M. Purica, D. Cristea, I. Cernica, R. Muller and V.
Moagar Poladian. Optimization of front surface texturing processes for high-
efficiency silicon solar cells. Solar Energy Materials and Solar Cells 87(1-4), pp.
423-431. 2005. . DOI: DOI: 10.1016/j.solmat.2004.06.013.
[31] M. Abbott and J. Cotter. Optical and electrical properties of laser texturing for
high-efficiency solar cells. Progress in Photovoltaics: Research and Applications
14pp. 225. 2006.
[32] R. Hendel. Laser applications in solar cell manufacturing. Laser Technik
Journal 5pp. 32. 2008.
[33] H. J. Booth. Recent applications of pulsed lasers in advanced materials
processing. Thin Solid Films, 453-454pp. 450-457. 2004.
[34] W. T. Silfvast. Laser Fundamentals 2000.
[35] J. Zhao, B. Huettner and A. Menschig. Microablation with ultrashort laser
pulses. Optics & Laser Technology 33(7), pp. 487-491. 2001. . DOI: DOI:
10.1016/S0030-3992(01)00066-4.
[36] C. Molpeceres, S. Lauzurica, J. J. García-Ballesteros, M. Morales, G. Guadaño,
J. L. Ocaña, S. Fernández, J. J. Gandía, F. Villar, O. Nos and J. Bertomeu. Selective
ablation of photovoltaic materials with UV laser sources for monolithic
interconnection of devices based on a-si:H. Materials Science and Engineering: B In
Press, Corrected Proof. DOI: DOI: 10.1016/j.mseb.2008.09.012.
136
[37] S. Rim, S. Zhao, S. R. Scully, M. D. McGehee and P. Peumans. An effective
light trapping configuration for thin-film solar cells. Appl. Phys. Lett. 91(24), pp.
243501. 2007. Available: http://link.aip.org/link/?APL/91/243501/1;
10.1063/1.2789677.
[38] J. Kang, C. Cho and J. Lee. Design of asymmetrically textured structure for
efficient light trapping in building-integrated photovoltaics. Organic Electronics
26pp. 61-65. 2015. . DOI: http://dx.doi.org/10.1016/j.orgel.2015.07.021.
[39] W. M. Steen. Laser Material Processing 1991.
[40] H. Klank, J. P. Kutter and O. Geschke. CO 2-laser micromachining and back-
end processing for rapid production of PMMA-based microfluidic systems. Lab on a
Chip 2(4), pp. 242-246. 2002.
[41] D. Snakenborg, H. Klank and J. P. Kutter. Microstructure fabrication with a
Co2 laser system. J Micromech Microengineering 14(2), pp. 182-189. 2004.
Available: http://dx.doi.org/10.1088/0960-1317/14/2/003.
[42] S. J. Qin and W. J. Li. Process characterization of fabricating 3D micro channel
systems by laser-micromachining. Sensors and Actuators A: Physical 97-98pp. 749-
757. 2002. . DOI: DOI: 10.1016/S0924-4247(02)00016-X.
[43] G. A. J. Markillie, H. J. Baker, F. J. Villarreal and D. R. Hall. Effect of
vaporization and melt ejection on laser machining of silica glass micro-optical
components. Appl. Opt. 41(27), pp. 5660-5667. 2002.
[44] A. Issa, D. Brabazon and M. S. J. Hashmi. 3D transient thermal modelling of
laser microchannel fabrication in lime-soda glass. J. Mater. Process. Technol. 207(1-
3), pp. 307-314. 2008. . DOI: DOI: 10.1016/j.jmatprotec.2008.06.056.
[45] S. Nikumb, Q. Chen, C. Li, H. Reshef, H. Y. Zheng, H. Qiu and D. Low.
Precision glass machining, drilling and profile cutting by short pulse lasers. Thin
Solid Films, 477(1-2), pp. 216-221. 2005.
[46] P. Kazansky, Y. Shimotsuma, J. Qiu, E. Bricchi and K. Hirao. "Nanostructuring
of transparent materials by light," in bragg gratings, photosensitivity, and poling in
glass waveguides. Technical Digest, Optical Society of Amercia Paper WA52003.
[47] T. H. R. Crawford and H. K. Haugen. Sub-wavelength surface structures on
silicon irradiated by femtosecond laser pulses at 1300 and 2100 nm wavelengths.
Appl. Surf. Sci. 253(11), pp. 4970-4977. 2007. . DOI: DOI:
10.1016/j.apsusc.2006.11.004.
[48] Hsiao-Yen Chung, Chiun-Hsun Chen, and Hsin-Sen Chu, "Analysis of
Pyramidal Surface Texturization of Silicon Solar Cells by Molecular Dynamics
Simulations," International Journal of Photoenergy, vol. vol. 2008, pp. 6 pages,
2008.
137
[49] P. Campbell and M. A. Green, "Light trapping properties of pyramidally
textured surfaces," Journal of Applied Physics, vol. 62, pp. 243-249, 1987.
[50] P. Campbell and M. A. Green. High performance light trapping textures for
monocrystalline silicon solar cells. Solar Energy Materials and Solar Cells 65(1-4),
pp. 369-375. 2001. . DOI: DOI: 10.1016/S0927-0248(00)00115-X.
[51] P. J. Sánchez-Illescas, P. Carpena, P. Bernaola-Galván, M. Sidrach-de-Cardona,
A. V. Coronado and J. L. Álvarez. An analysis of geometrical shapes for PV module
glass encapsulation. Solar Energy Mater. Solar Cells 92(3), pp. 323-331. 2008. .
DOI: DOI: 10.1016/j.solmat.2007.09.008.
[52] J. Deubener, G. Helsch, A. Moiseev and H. Bornhöft. Glasses for solar energy
conversion systems. Journal of the European Ceramic Society In Press, Corrected
Proof2008. . DOI: DOI: 10.1016/j.jeurceramsoc.2008.08.009.
[53] M. Callies, Y. Chen, F. Marty, A. Pépin and D. Quéré. Microfabricated textured
surfaces for super-hydrophobicity investigations. Microelectronic Engineering 78-
79pp. 100-105. 2005. . DOI: DOI: 10.1016/j.mee.2004.12.093.
[54] M. A. Green, P. A. Basore, N. Chang, D. Clugston, R. Egan, R. Evans, D.
Hogg, S. Jarnason, M. Keevers, P. Lasswell, J. O’Sullivan, U. Schubert, A. Turner,
S. R. Wenham and T. Young. Crystalline silicon on glass (CSG) thin-film solar cell
modules. Solar Energy 77(6), pp. 857-863. 2004. . DOI: DOI:
10.1016/j.solener.2004.06.023.
[55] J. Muller, G. Schope, B. Rech, H. Schade, P. Lechner, R. Geyer, H. Stiebig and
W. Reetz. Role of the glass/TCO substrate in thin film silicon solar cells.
Photovoltaic Energy Conversion, 2003. Proceedings of 3rd World Conference On
2pp. 1839-1842 Vol.2. 2003.
[56] J. Müller, B. Rech, J. Springer and M. Vanecek. TCO and light trapping in
silicon thin film solar cells. Solar Energy 77(6), pp. 917-930. 2004. . DOI: DOI:
10.1016/j.solener.2004.03.015.
[57] A. J. Nozik. Quantum dot solar cells. Physica E: Low-Dimensional Systems and
Nanostructures 14(1-2), pp. 115-120. 2002. . DOI: DOI: 10.1016/S1386-
9477(02)00374-0.
[58] M. Dasog, K. Bader and J. G. C. Veinot. Influence of halides on the optical
properties of silicon quantum dots. Chem. Mater. 27(4), pp. 1153-1156. 2015.
Available: http://dx.doi.org/10.1021/acs.chemmater.5b00115. DOI:
10.1021/acs.chemmater.5b00115.
[59] L. H. Lie, M. Duerdin, E. M. Tuite, A. Houlton and B. R. Horrocks. Preparation
and characterisation of luminescent alkylated-silicon quantum dots. J Electroanal
Chem 538–539pp. 183-190. 2002. . DOI:
http://dx.doi.org.dcu.idm.oclc.org/10.1016/S0022-0728(02)00994-4.
138
[60] H. F. Wilson, L. McKenzie-Sell and A. S. Barnard. Shape dependence of the
band gaps in luminescent silicon quantum dots. J. Mater. Chem. C 2(44), pp. 9451-
9456. 2014. Available: http://dx.doi.org/10.1039/C4TC01312C. DOI:
10.1039/C4TC01312C.
[61] D. Kovalev, H. Heckler, G. Polisski, J. Diener and F. Koch. Optical properties
of silicon nanocrystals. Optical Materials 17(1-2), pp. 35-40. 2001. . DOI: DOI:
10.1016/S0925-3467(01)00017-9.
[62] P. M. Fauchet. Light emission from si quantum dots. Materials Today 8(1), pp.
26-33. 2005. . DOI: DOI: 10.1016/S1369-7021(04)00676-5.
[63] M. Rosso-Vasic, E. Spruijt, B. v. Lagen, L. D. Cola and H. Zuilhof. Alkyl-
functionalized oxide-free silicon nanoparticles: Synthesis and optical properties.
Small 4(10), pp. 1835-1841. 2008. Available:
http://dx.doi.org/10.1002/smll.200800066.
[64] P. Mishra and K. P. Jain. Raman, photoluminescence and optical absorption
studies on nanocrystalline silicon. Materials Science and Engineering B 95(3), pp.
202-213. 2002. . DOI: DOI: 10.1016/S0921-5107(02)00234-9.
[65] X. J. Hao, E. -. Cho, G. Scardera, E. Bellet-Amalric, D. Bellet, Y. S. Shen, S.
Huang, Y. D. Huang, G. Conibeer and M. A. Green. Effects of phosphorus doping
on structural and optical properties of silicon nanocrystals in a SiO2 matrix. Thin
Solid Films 517(19), pp. 5646-5652. 2009. . DOI: DOI: 10.1016/j.tsf.2009.02.076.
[66] S. Giménez, I. Mora-Seró, L. Macor, N. Guijarro, T. Lana-Villarreal, R. Gómez,
L. J. Diguna, Q. Shen, T. Toyoda and J. Bisquert. Improving the performance of
colloidal quantum-dot-sensitized solar cells. Nanotechnology 20(29), pp. 295204.
2009. Available: http://stacks.iop.org/0957-4484/20/i=29/a=295204.
[67] M. Samadpour, Z. Ghane, N. Ghazyani, F. Tajabadi and N. Taghavinia.
Monolithic quantum dot sensitized solar cells. J. Phys. D 46(48), pp. 485101. 2013.
Available: http://stacks.iop.org/0022-3727/46/i=48/a=485101.
[68] K. Kalyanasundaram. Dye-Sensitized Solar Cells 2010Available:
https://books.google.ie/books?id=1n1QC2snf-AC.
[69] T. Soga. Nanostructured Materials for Solar Energy Conversion, 1st Ed
2006Available: http://prism.talis.com/dcu/items/595708;
http://www.loc.gov/catdir/enhancements/fy0665/2006050753-d.html.
[70] A. Pattantyus-Abraham, I. J. Kramer, A. R. Barkhouse, X. Wang, G.
Konstantatos, R. Debnath, L. Levina, I. Raabe, M. K. Nazeeruddin, M. Grätzel and
E. H. Sargent. Depleted-heterojunction colloidal quantum dot solar cells. ACS Nano
4(6), pp. 3374-3380. 2010. Available: http://dx.doi.org/10.1021/nn100335g. DOI:
10.1021/nn100335g.
139
[71] R. A. Synowicki, B. D. Johs and A. C. Martin. Optical properties of soda-lime
float glass from spectroscopic ellipsometry. Thin Solid Films 519(9), pp. 2907-2913.
2011. . DOI: 10.1016/j.tsf.2010.12.110.
[72] H. Y. Zheng, W. Zhou, H. X. Qian, T. T. Tan and G. C. Lim. Polarisation-
independence of femtosecond laser machining of fused silica. Appl. Surf. Sci. 236(1-
4), pp. 114-119. 2004. . DOI: DOI: 10.1016/j.apsusc.2004.04.034.
[73] L. Gallais, P. Cormont and J. Rullier. Investigation of stress induced by CO2
laserprocessing of fused silica optics for laser damagegrowth mitigation.
Opt.Express 17(26), pp. 23488-23501. 2009. Available:
http://www.opticsexpress.org/abstract.cfm?URI=oe-17-26-23488.
[74] T. L. H Y Zheng and. Studies of CO 2 laser peeling of glass substrates. J
Micromech Microengineering 15(11), pp. 2093. 2005. Available:
http://stacks.iop.org/0960-1317/15/i=11/a=014.
[75] C. Iliescu, B. Chen and J. Miao. On the wet etching of pyrex glass. Sensors and
Actuators A: Physical 143(1), pp. 154-161. 2008. . DOI: DOI:
10.1016/j.sna.2007.11.022.
[76] J. Wong, J. L. Ferriera, E. F. Lindsey, D. L. Haupt, I. D. Hutcheon and J. H.
Kinney. Morphology and microstructure in fused silica induced by high fluence
ultraviolet 3ω (355 nm) laser pulses. Journal of Non-Crystalline Solids 352(3), pp.
255-272. 2006. . DOI: DOI: 10.1016/j.jnoncrysol.2005.11.036.
[77] K. L. Chopra and S. R. Das. Thin Film Solar Cells 1983.
[78] D. Mukherjee and S. Chakrabarti. Fundamentals of Renewable Energy Systems
2004.
[79] J. Petzing, J. Coupland and R. Leach. Good practice guide no. 116 - the
measurement of rough surface topography using coherence scanning interferometry.
Npl (116), 2010.
[80] R. K. Leach. Optical Measurement of Surface Topography 2011.
[81] A. Issa, "Computational Control of Laser Systems for micro-machining," vol. 1,
2007.
[82] S. K. Nayar. Shape from focus. Robotics Institute. Pittsburgh, PA,. 1989.
[83] e. a. F Marinello. Development and analysis of a software tool for stitching
three-dimensional surface topography data sets. Measurement Science and
Technology 18(5), pp. 1404. 2007. Available: http://stacks.iop.org/0957-
0233/18/i=5/a=028.
[84] e. a. F Marinello. Increase of maximum detectable slope with optical profilers,
through controlled tilting and image processing. Measurement Science and
140
Technology 18(2), pp. 384. 2007. Available: http://stacks.iop.org/0957-
0233/18/i=2/a=S09.
[85] L. Fan, F. Song and S. Jutamulia. Edge detection with large depth of focus
using differential Haar–Gaussian wavelet transform. Opt. Commun. 270(2), pp. 169-
175. 2007. . DOI: DOI: 10.1016/j.optcom.2006.09.015.
[86] P. Grossmann. Depth from focus. Pattern Recog. Lett. 5(1), pp. 63-69. 1987. .
DOI: DOI: 10.1016/0167-8655(87)90026-2.
[87] M. F. Hansen, G. A. Atkinson, L. N. Smith and M. L. Smith. 3D face
reconstructions from photometric stereo using near infrared and visible light.
Comput. Vision Image Understanding 114(8), pp. 942-951. 2010. . DOI: DOI:
10.1016/j.cviu.2010.03.001.
[88] F. S. Helmli and S. Scherer. Adaptive shape from focus with an error estimation
in light microscopy. Presented at Image and Signal Processing and Analysis, 2001.
ISPA 2001. Proceedings of the 2nd International Symposium On. 2001, .
[89] M. Hocaoglu and M. Unel. HK segmentation of 3D micro-structures
reconstructed from focus. 5226pp. 1173-1180. 2008. . DOI: 10.1007/978-3-540-
87442-3_145.
[90] W. Huang and Z. Jing. Evaluation of focus measures in multi-focus image
fusion. Pattern Recog. Lett. 28(4), pp. 493-500. 2007. . DOI: DOI:
10.1016/j.patrec.2006.09.005.
[91] A. S. Malik, Seong-O Shim and Tae-Sun Choi. Depth map estimation using a
robust focus measure. Presented at Image Processing, 2007. ICIP 2007. IEEE
International Conference On. 2007, .
[92] A. S. Malik and Tae-Sun Choi. Analysis of effects of texture reflectance and
source illumination on focus measures for microscopic images. Presented at
Computer and Electrical Engineering, 2009. ICCEE '09. Second International
Conference On. 2009, .
[93] A. S. Malik and T. Choi. Consideration of illumination effects and optimization
of window size for accurate calculation of depth map for 3D shape recovery. Pattern
Recognit 40(1), pp. 154-170. 2007. . DOI: DOI: 10.1016/j.patcog.2006.05.032.
[94] A. S. Malik and T. Choi. A novel algorithm for estimation of depth map using
image focus for 3D shape recovery in the presence of noise. Pattern Recognit 41(7),
pp. 2200-2225. 2008. . DOI: DOI: 10.1016/j.patcog.2007.12.014.
[95] M. Niederöst, J. Niederöst and J. Skucka. Shape from focus fully automated 3D
reconstruction and visualization of microscopic objects.
141
[96] M. Noguchi and S. K. Nayar. Microscopic shape from focus using a projected
illumination pattern, Math. Comput. Model. 24(5-6), pp. 31-48. 1996. . DOI: DOI:
10.1016/0895-7177(96)00114-8.
[97] S. Shim and T. Choi. Depth from focus based on combinatorial optimization.
Opt. Lett. 35(12), pp. 1956-1958. 2010. Available:
http://ol.osa.org/abstract.cfm?URI=ol-35-12-1956.
[98] A. G. Valdecasas, D. Marshall, J. M. Becerra and J. J. Terrero. On the extended
depth of focus algorithms for bright field microscopy. Micron 32(6), pp. 559-569.
2001. . DOI: DOI: 10.1016/S0968-4328(00)00061-5.
[99] J. Vitrià and J. Llacer. Reconstructing 3D light microscopic images using the
EM algorithm. Pattern Recog. Lett. 17(14), pp. 1491-1498. 1996. . DOI: DOI:
10.1016/S0167-8655(96)00104-3.
[100] T. V. Vorburger and E. C. Teague. Optical techniques for on-line measurement
of surface topography. Precis Eng 3(2), pp. 61-83. 1981. . DOI: DOI: 10.1016/0141-
6359(81)90038-6.
[101] Y. Wu and C. Wang. Extended depth of focus image for phytolith analysis.
Journal of Archaeological Science 36(10), pp. 2253-2257. 2009. . DOI: DOI:
10.1016/j.jas.2009.06.010.
[102] Yu Song, Mantian Li and Lining Sun. A new auto-focusing algorithm for
optical microscope based automated system. Presented at Control, Automation,
Robotics and Vision, 2006. ICARCV '06. 9th International Conference On. 2006, .
[103] Yu Sun, S. Duthaler and B. J. Nelson. Autofocusing algorithm selection in
computer microscopy. Presented at Intelligent Robots and Systems, 2005. (IROS
2005). 2005 IEEE/RSJ International Conference On. 2005, .
[104] J. C. Russ. The Image Processing Handbook (6th ed. ed.) 2011.
[105] V. K. Narasimhan and C. Yi. Nanostructures for photon management in solar
cells. Nanophotonics 2pp. 187. 2013. . DOI: 10.1515/nanoph-2013-0001.
[106] International Standards Organisation 4287, "Geometrical Product
Specifications (GPS) -- Surface texture: Profile method -- Terms, definitions and
surface texture parameters," 1997.
[107] International Standards Organisation 4288, "Geometrical Product
Specifications (GPS) -- Surface texture: Profile method -- Rules and procedures for
the assessment of surface texture," 1996.
[108] International Standards Organisation 230-2, "Test code for machine tools- Part
2: Determination of accuracy and repeatability of positioning numerically controlled
axes." 2006.
142
[109] D. J. Whitehouse. Surfaces and their Measurement 2002.
[110] D. G. Chetwynd, J. A. Greenwood and E. Mainsah. Metrology and Properties
of Engineering Surfaces 2001.
[111] W. Sun and J. D. Claverley. Verification of an optical micro-CMM using the
focus variation technique: Aspects of probing errors. CIRP Ann. Manuf. Technol.
64(1), pp. 511-514. 2015. . DOI: http://dx.doi.org/10.1016/j.cirp.2015.04.089.
[112] K. J. Gåsvik. Optical Metrology (3rd ed. ed.) 2002.
[113] Rofin-Baasel Ltd, "Rofin DC-015 CO2 laser operation manuals," 2004.
[114] S. L. Storm and A. Springsteen, "Choosing the Right Sphere Size
for Your Application," .
[115] D. Moore, S. Krishnamurthy, Y. Chao, Q. Wang, D. Brabazon and P. J.
McNally. Characteristics of silicon nanocrystals for photovoltaic applications.
Physica Status Solidi (a) 208(3), pp. 604-607. 2011. Available:
http://dx.doi.org/10.1002/pssa.201000381. DOI: 10.1002/pssa.201000381.
[116] Y. Chao, S. Krishnamurthy, M. Montalti, L. H. Lie, A. Houlton, B. R.
Horrocks, L. Kjeldgaard, V. R. Dhanak, M. R. C. Hunt and L. Šiller. Reactions and
luminescence in passivated si nanocrystallites induced by vacuum ultraviolet and
soft-x-ray photons. J. Appl. Phys. 98(4), pp. 044316. 2005. . DOI:
http://dx.doi.org/10.1063/1.2012511.
[117] Y. Chao, L. Šiller, S. Krishnamurthy, P. R. Coxon, U. Bangert, M. Gass, L.
Kjeldgaard, S. N. Patole, L. H. Lie, N. O'Farrell, T. A. Alsop, A. Houlton and B. R.
Horrocks. Evaporation and deposition of alkyl-capped silicon nanocrystals in
ultrahigh vacuum. Nature Nanotechnology (8), pp. 486-489. 2007. . DOI:
10.1038/nnano.2007.224.
[118] Y. Chao, A. Houlton, B. R. Horrocks, M. R. C. Hunt, N. R. J. Poolton, J. Yang
and L. Siller. Optical luminescence from alkyl-passivated si nanocrystals under
vacuum ultraviolet excitation: Origin and temperature dependence of the blue and
orange emissions. Appl. Phys. Lett. 88(26), pp. 263119. 2006. Available:
http://link.aip.org/link/?APL/88/263119/1. DOI: 10.1063/1.2216911.
[119] E. Duffy, X. He, E. P. Nesterenko, D. Brabazon, A. Dey, S. Krishnamurthy, P.
N. Nesterenko and B. Paull. Thermally controlled growth of carbon onions within
porous graphitic carbon-detonation nanodiamond monolithic composites. RSC Adv.
(29), pp. 22906. 2015. . DOI: 10.1039/C5RA00258C.
[120] X. He, K. B. Male, P. N. Nesterenko, D. Brabazon, B. Paull and J. H. T.
Luong. Adsorption and desorption of methylene blue on porous carbon monoliths
and nanocrystalline cellulose. ACS Appl. Mater. Interfaces (17), pp. 8796. 2013. .
DOI: 10.1021/am403222u.
143
[121] X. He, E. P. Nesterenko, P. N. Nesterenko, D. Brabazon, L. Zhou, J. D.
Glennon, J. H. T. Luong and B. Paull. Fabrication and characterization of
nanotemplated carbon monolithic material. ACS Appl. Mater. Interfaces (17), pp.
8572. 2013. . DOI: 10.1021/am402030m.
[122] M. Vazquez, D. Moore, X. He, A. Ben Azouz, E. Nesterenko, P. Nesterenko,
B. Paull and D. Brabazon. Focussed ion beam serial sectioning and imaging of
monolithic materials for 3D reconstruction and morphological parameter evaluation.
Analyst (1), pp. 99. 2013. . DOI: 10.1039/C3AN01827J.
[123] A. H. Eltmimi, L. Barron, A. Rafferty, J. P. Hanrahan, O. Fedyanina, E.
Nesterenko, P. N. Nesterenko and B. Paull. Preparation, characterisation and
modification of carbon-based monolithic rods for chromatographic applications.
Journal of Separation Science 33(9), pp. 1231-1243. 2010. . DOI:
10.1002/jssc.200900845.
[124] H. Giesche. Mercury porosimetry: A general (practical) overview. Particle &
Particle Systems Characterization 23(1), pp. 9-19. 2006. . DOI:
10.1002/ppsc.200601009.
[125] S. M. Karazi, A. Issa and D. Brabazon. Comparison of ANN and DoE for the
prediction of laser-machined micro-channel dimensions. Optics and Lasers in
Engineering 47(9), pp. 956-964. 2009. Available:
http://www.sciencedirect.com/science/article/pii/S0143816609000943.
[126] K. Jäger, O. Isabella, L. Zhao and M. Zeman. Light scattering properties of
surface-textured substrates. Physica Status Solidi (C) 7(3-4), pp. 945-948. 2010.
Available: http://dx.doi.org/10.1002/pssc.200982695.
[127] M. Kolli, M. Hamidouche, N. Bouaouadja and G. Fantozzi. HF etching effect
on sandblasted soda-lime glass properties. Journal of the European Ceramic Society
29(13), pp. 2697-2704. 2009. . DOI: DOI: 10.1016/j.jeurceramsoc.2009.03.020.
[128] J. Zhao, J. Sullivan and T. D. Bennett. Wet etching study of silica glass after
CW CO2 laser treatment. Applied Surface Science 225(1-4), pp. 250-255. 2004. .
DOI: DOI: 10.1016/j.apsusc.2003.10.012.
[129] C. Yang, R. A. Bley, S. M. Kauzlarich, H. W. H. Lee and G. R. Delgado.
Synthesis of alkyl-terminated silicon nanoclusters by a solution route. J. Am. Chem.
Soc. 121(22), pp. 5191-5195. 1999. Available: http://dx.doi.org/10.1021/ja9828509.
DOI: 10.1021/ja9828509.
[130] I. Vasiliev, J. R. Chelikowsky and R. M. Martin. Surface oxidation effects on
the optical properties of silicon nanocrystals. Phys.Rev.B 65(12), pp. 121302. 2002. .
DOI: 10.1103/PhysRevB.65.121302.
[131] A. P. Cocco, G. J. H. Nelson W.M., A. Nakajo, T. D. Myles, A. M. Kiss, J. J.
Lombardo and W. K. S. Chiu. Three-dimensional microstructural imaging methods
144
for energy materials. Phys. Chem. Chem. Phys. (39), pp. 16377. 2013. . DOI:
10.1039/C3CP52356J.
[132] Design-Expert software, "“v9, user’s guide, Technical manual”, Stat-Ease
Inc., ," Stat-Ease Inc., 2015.
i
Appendix A : Positional Analysis
The following tables contain the results and calculations performed in analysing the
performance characteristics of the stage of the microscope.
ii
Table 32: Large scale positional analysis of X axis using Mahr 810SW dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
+ - + - + - + - + -
Target Position (mm) 0.48 1.09 1.53 1.76 2.75
Positional Deviations
(µm)
1 0 10 -10 0 -10 -10 -10 -10 -20 -10
2 0 10 -10 10 -10 -10 -10 -10 -20 -10
3 0 10 0 0 -10 -10 -10 -10 -20 -10
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
0 10 -10 0 -10 -10 -10 -10 -20 -10
Estimator of standard
uncertainty 𝒔𝒊 (µm) 0 0 10 10 0 0 0 0 0 0
𝟐𝒔𝒊 0 0 20 20 0 0 0 0 0 0
�̅�𝒊 − 𝟐𝒔𝒊 0 10 -30 -20 -10 -10 -10 -10 -20 -10
�̅�𝒊 + 𝟐𝒔𝒊 0 10 10 20 -10 -10 -10 -10 -20 -10
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
0 0 40 40 0 0 0 0 0 0
Reverse value 𝑩𝒊 (µm) -10 -10 0 0 -10
Bidirectional
repeatability 𝑹𝒊 (µm) 10 80 0 0 10
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
10 -10 -10 -10 -20
iii
Table 33: Large scale positional analysis of X axis using Mahr 810SW dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction + - + - + - + - + -
Target Position (mm) 2.88 3.08 3.72 4.95 5.1
Positional Deviations
(µm)
1 -20 -20 -10 -10 -20 -20 -30 -20 -20 -20
2 -20 -20 -10 10 -20 0 -30 -10 -30 -30
3 -20 -10 -10 -10 -20 -10 -20 -20 -30 -20
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
-20 -20 -10 0 -20 -10 -30 -20 -30 -20
Estimator of standard
uncertainty 𝒔𝒊 (µm)
0 10 0 10 0 10 10 10 10 10
𝟐𝒔𝒊 0 20 0 20 0 20 20 20 20 20
�̅�𝒊 − 𝟐𝒔𝒊 -20 -40 -10 -20 -20 -30 -50 -40 -50 -40
�̅�𝒊 + 𝟐𝒔𝒊 -20 0 -10 20 -20 10 -10 0 -10 0
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
0 40 0 40 0 40 40 40 40 40
Reverse value 𝑩𝒊 (µm) 0 -10 -10 -10 -10
Bidirectional
repeatability 𝑹𝒊 (µm) 60 60 60 80 80
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
-20 -10 -20 -30 -30
iv
Table 34: Large scale positional analysis of Y axis using Mahr 810SW dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
+ - + - + - + - + -
Target Position (mm) 0.48 1.09 1.53 1.76 2.75
Positional Deviations
(µm)
1 -10 -10 -10 0 -10 -10 -10 0 -10 -10
2 -10 10 -20 0 -20 0 -20 0 0 -10
3 -20 0 -20 0 -20 0 -20 -10 -20 -10
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
-10 0 -20 0 -20 0 -20 0 -10 -10
Estimator of standard
uncertainty 𝒔𝒊 (µm)
10 10 10 0 10 10 10 10 10 0
𝟐𝒔𝒊 20 20 20 0 20 20 20 20 20 0
�̅�𝒊 − 𝟐𝒔𝒊 -30 -20 -40 0 -40 -20 -40 -20 -30 -10
�̅�𝒊 + 𝟐𝒔𝒊 10 20 0 0 0 20 0 20 10 -10
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
40 40 40 0 40 40 40 40 40 0
Reverse value 𝑩𝒊 (µm) -10 -20 -20 -20 0
Bidirectional
repeatability 𝑹𝒊 (µm) 80 60 80 80 60
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
-10 -10 -10 -10 -10
v
Table 35: Large scale positional analysis of Y axis using Mahr 810SW dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction + - + - + - + - + -
Target Position (mm) 2.88 3.08 3.72 4.95 5.1
Positional Deviations
(µm)
1 -20 -10 -10 -10 -20 -10 -20 -20 -20 -20
2 -10 -10 0 0 -10 -10 -20 -10 -20 -10
3 -30 -10 -30 -10 -30 -20 -30 -20 -30 -20
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
-20 -10 -10 -10 -20 -10 -20 -20 -20 -20
Estimator of standard
uncertainty 𝒔𝒊 (µm)
10 0 20 10 10 10 10 10 10 10
𝟐𝒔𝒊 20 0 40 20 20 20 20 20 20 20
�̅�𝒊 − 𝟐𝒔𝒊 -40 -10 -50 -30 -40 -30 -40 -40 -40 -40
�̅�𝒊 + 𝟐𝒔𝒊 0 -10 30 10 0 10 0 0 0 0
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
40 0 80 40 40 40 40 40 40 40
Reverse value 𝑩𝒊 (µm) -10 0 -10 0 0
Bidirectional
repeatability 𝑹𝒊 (µm) 60 140 80 80 80
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
-20 -10 -20 -20 -20
vi
Table 36: Large scale positional analysis of Z axis using Mahr 810SW dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
+ - + - + - + - + -
Target Position (mm) 0.05 0.11 0.15 0.18 0.28
Positional Deviations
(µm)
1 0 0 0 0 0 10 0 0 0 0
2 0 0 0 0 0 10 -10 0 -10 0
3 0 0 0 0 10 10 -10 0 -10 0
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
0 0 0 0 0 10 -10 0 -10 0
Estimator of standard
uncertainty 𝒔𝒊 (µm)
0 0 0 0 10 0 10 0 10 0
𝟐𝒔𝒊 0 0 0 0 20 0 20 0 20 0
�̅�𝒊 − 𝟐𝒔𝒊 0 0 0 0 -20 10 -30 0 -30 0
�̅�𝒊 + 𝟐𝒔𝒊 0 0 0 0 20 10 10 0 10 0
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
0 0 0 0 40 0 40 0 40 0
Reverse value 𝑩𝒊 (µm) 0 0 -10 -10 -10
Bidirectional
repeatability 𝑹𝒊 (µm) 0 0 60 60 60
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
0 0 10 -10 -10
vii
Table 37: Large scale positional analysis of Z axis using Mahr 810SW dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction + - + - + - + - + -
Target Position (mm) 0.29 0.31 0.38 0.50 0.51
Positional Deviations
(µm)
1 0 0 10 0 10 0 10 10 10 10
2 0 0 0 0 0 10 0 10 10 10
3 0 0 0 0 0 0 10 0 0 10
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
0 0 0 0 0 0 10 10 10 10
Estimator of standard
uncertainty 𝒔𝒊 (µm) 0 0 10 0 10 10 10 10 10 0
𝟐𝒔𝒊 0 0 20 0 20 20 20 20 20 0
�̅�𝒊 − 𝟐𝒔𝒊 0 0 -20 0 -20 -20 -10 -10 -10 10
�̅�𝒊 + 𝟐𝒔𝒊 0 0 20 0 20 20 30 30 30 10
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
0 0 40 0 40 40 40 40 40 0
Reverse value 𝑩𝒊 (µm) 0 0 0 0 0
Bidirectional
repeatability 𝑹𝒊 (µm) 0 60 80 80 60
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
0 0 0 10 10
viii
Table 38: Small scale positional analysis of X axis using Mitutoyo MT513-405E dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
Target Position (mm) 0.020 0.045 0.060 0.070 0.110
Positional Deviations
(µm)
1 0 0 -1 -2 1 1 0 1 2 2
2 1 2 -1 1 2 4 2 4 3 4
3 0 2 -1 0 2 3 2 3 2 4
Mean unidirectional
positional deviation �̅�𝒊
(µm)
0 1 -1 0 2 3 1 3 2 3
Estimator of standard
uncertainty 𝒔𝒊 (µm) 1 1 0 2 1 2 1 2 1 1
𝟐𝒔𝒊 2 2 0 4 2 4 2 4 2 2
�̅�𝒊 − 𝟐𝒔𝒊 -2 -1 -1 -4 0 -1 -1 -1 0 1
�̅�𝒊 + 𝟐𝒔𝒊 2 3 -1 4 4 7 3 7 4 5
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
4 4 0 8 4 8 4 8 4 4
Reverse value 𝑩𝒊 (µm) -1 -1 -1 -2 -1
Bidirectional
repeatability 𝑹𝒊 (µm) 8 12 14 14 8
Mean bidirectional
positional deviation �̅�𝒊
(µm)
1 -1 3 2 3
ix
Table 39: Small scale positional analysis of X axis using Mitutoyo MT513-405E dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓
Target Position (mm) 0.115 0.125 0.150 0.200 0.205
Positional Deviations (µm)
1 0 -1 -1 -1 1 1 2 2 -1 -1
2 1 1 0 1 3 4 4 4 1 1
3 0 1 -1 1 2 3 3 4 1 0
Mean unidirectional
positional deviation 𝒙𝒊 (µm) 0 0 -1 0 2 3 3 3 0 0
Estimator of standard
uncertainty 𝒔𝒊 (µm) 1 1 1 1 1 2 1 1 1 1
𝟐𝒔𝒊 2 2 2 2 2 4 2 2 2 2
�̅�𝒊 − 𝟐𝒔𝒊 -2 -2 -3 -2 0 -1 1 1 -2 -2
�̅�𝒊 + 𝟐𝒔𝒊 2 2 1 2 4 7 5 5 2 2
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊 (µm) 4 4 4 4 4 8 4 4 4 4
Reverse value 𝑩𝒊 (µm) 0 -1 -1 0 0
Bidirectional repeatability
𝑹𝒊 (µm) 8 8 14 8 8
Mean bidirectional
positional deviation 𝒙𝒊 (µm) 0 -1 3 3 0
x
Table 40: Small scale positional analysis of Y axis using Mitutoyo MT513-405E dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
+ - + - + - + - + -
Target Position (mm) 0.020 0.045 0.060 0.070 0.110
Positional Deviations
(µm)
1 -2 1 -2 1 -2 2 -1 2 -1 2
2 1 6 1 6 2 6 2 7 3 8
3 0 4 1 5 1 6 2 6 3 -3
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
0 4 0 4 0 5 1 5 2 2
Estimator of standard
uncertainty 𝒔𝒊 (µm) 2 3 2 3 2 2 2 3 2 6
𝟐𝒔𝒊 4 6 4 6 4 4 4 6 4 12
�̅�𝒊 − 𝟐𝒔𝒊 -4 -2 -4 -2 -4 1 -3 -1 -2 -10
�̅�𝒊 + 𝟐𝒔𝒊 4 10 4 10 4 9 5 11 6 14
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
8 12 8 12 8 8 8 12 8 24
Reverse value 𝑩𝒊 (µm) -4 -4 -5 -4 0
Bidirectional repeatability
𝑹𝒊 (µm) 22 22 16 22 40
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
2 2 3 3 2
xi
Table 41: Small scale positional analysis of Y axis using Mitutoyo MT513-405E dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction + - + - + - + - + -
Target Position (mm) 0.115 0.125 0.150 0.200 0.205
Positional Deviations (µm)
1 -1 2 -1 2 0 3 2 5 2 3
2 3 7 3 7 4 8 6 9 7 7
3 3 6 3 5 4 6 5 8 5 7
Mean unidirectional
positional deviation 𝒙𝒊 (µm) 2 5 2 5 3 6 4 7 5 6
Estimator of standard
uncertainty 𝒔𝒊 (µm) 2 3 2 3 2 3 2 2 3 2
𝟐𝒔𝒊 4 6 4 6 4 6 4 4 6 4
�̅�𝒊 − 𝟐𝒔𝒊 -2 -1 -2 -1 -1 0 0 3 -1 2
�̅�𝒊 + 𝟐𝒔𝒊 6 11 6 11 7 12 8 11 11 10
Unidirectional Repeatability
𝑹𝒊 = 𝟒𝒔𝒊 (µm) 8 12 8 12 8 12 8 8 12 8
Reverse value 𝑩𝒊 (µm) -3 -3 -3 -3 -1
Bidirectional repeatability
𝑹𝒊 (µm) 22 22 22 16 22
Mean bidirectional
positional deviation 𝒙𝒊 (µm) 4 4 5 6 6
xii
Table 42: Small scale positional analysis of Z axis using Mitutoyo MT513-405E dial gauge indicator (1-5)
Position Number i 1 2 3 4 5
Approach Direction
+ - + - + - + - + -
Target Position (mm) 0.019 0.044 0.061 0.070 0.110
Positional Deviations
(µm)
1 0 11 1 11 1 12 0 11 1 11
2 3 13 2 12 2 13 2 12 2 12
3 -1 8 -1 8 0 8 -1 8 -1 8
Mean unidirectional
positional deviation 𝒙𝒊
(µm)
1 11 1 10 1 11 0 10 1 10
Estimator of standard
uncertainty 𝒔𝒊 (µm) 2 3 2 2 1 3 2 2 2 2
𝟐𝒔𝒊 4 6 4 4 2 6 4 4 4 4
�̅�𝒊 − 𝟐𝒔𝒊 -3 5 -3 6 -1 5 -4 6 -3 6
�̅�𝒊 + 𝟐𝒔𝒊 5 17 5 14 3 17 4 14 5 14
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊
(µm)
8 12 8 8 4 12 8 8 8 8
Reverse value 𝑩𝒊 (µm) -10 -9 -10 -10 -9
Bidirectional
repeatability 𝑹𝒊 (µm) 22 17 20 18 17
Mean bidirectional
positional deviation 𝒙𝒊
(µm)
6 6 6 5 6
xiii
Table 43: Small scale positional analysis of Z axis using Mitutoyo MT513-405E dial gauge indicator (6-10)
Position Number i 6 7 8 9 10
Approach Direction + - + - + - + - + -
Target Position (mm) 0.115 0.123 0.149 0.198 0.204
Positional Deviations (µm)
1 0 10 0 11 0 10 0 10 1 9
2 1 13 1 11 1 11 0 11 0 10
3 -1 8 -3 8 -1 7 -2 7 -3 6
Mean unidirectional
positional deviation 𝒙𝒊 (µm)
0 10 -1 10 0 9 -1 9 -1 8
Estimator of standard
uncertainty 𝒔𝒊 (µm)
1 3 2 2 1 2 1 2 2 2
𝟐𝒔𝒊 2 6 4 4 2 4 2 4 4 4
�̅�𝒊 − 𝟐𝒔𝒊 -2 4 -5 6 -2 5 -3 5 -5 4
�̅�𝒊 + 𝟐𝒔𝒊 2 16 3 14 2 13 1 13 3 12
Unidirectional
Repeatability 𝑹𝒊 = 𝟒𝒔𝒊 (µm)
4 12 8 8 4 8 4 8 8 8
Reverse value 𝑩𝒊 (µm) -10 -11 -9 -10 -9
Bidirectional repeatability
𝑹𝒊 (µm)
20 19 15 16 17
Mean bidirectional
positional deviation 𝒙𝒊 (µm)
5 5 5 4 4
xiv
Appendix B : RSM Procedure
To produce and evaluate a RSM to model on the effect of laser micromachining
parameters on the parasitic absorption loss in the textured superstrates, the following
steps need to be performed:
The identification of the key process input variables.
The key factors can be defined from the previous literature or by conducting a
screening study. In the case of the work completed in this thesis, a combination of both
was completed. A previous study had worked using the same laser micromachining
system for the micro-ablation of channels in a different glass material [81]. Using this
information as a starting point, the process variables were further investigated through
a screening study to ensure compatibility with the materials used in this study. The
process input factors were laser power (Duty Cycle), pulse repetition frequency (PRF),
and translation speed (U).
The determination of each input ranges.
The screening study also helped finding the ranges of each factor. The criterion of
selecting the working ranges was getting a smooth, uniform, and continuous micro-
channel.
xv
Testing the significance of the model terms
In order to ensure that the model terms are significant, an analysis of the following
terms was completed [132].
The F value is calculated by term mean square divided by residual mean square.
The p-value shows whether the model/term is significant or not. The p-value is the
probability of getting an F Value of this size if the model/term did not have an effect
on the response. In general, a term that has a probability value less than 0.05, which is
the significance threshold that sets the risk level in significance/non-significance
detection, would be considered a significant effect.
The R-Squared is a measure of the amount of variation around the mean explained
by the model. A low R-Squared value indicates variation around the average
predictions of the model.
The Adj R-Squared is a measure of the amount of variation around the mean
explained by the model, adjusted for the number of terms in the model. The adjusted
R-Squared plateaus as the number of terms in the model increases if those additional
terms do not add value to the model.
The Pred R-Squared is a measure of the amount of variation in new data explained
by the model. This value decreases when too many insignificant terms are added to
the model.
The Adequate Precision is a signal to noise ratio. It compares the range of the
predicted values at the design points to the average prediction error. This value should
be larger than 4 to indicate that the model signal is strong.
The automatic Stepwise regression method was utilised in the Design Expert software
to eliminate insignificant terms in this work.
xvi
Appendix C : Dial Gauge Technical
Specifications
Mahr MarCator 810W / SW Dial Gauge Specifications
xvii
Mitutoyo 513-405E Dial Gauge Specifications
xviii
Appendix D : ISO 230-2
Definitions
Target position - 𝑃𝑖 (i = 1 to m): The position to which the moving part is commanded
to move. i identifies the particular position among other selected target positions along
the axis.
Actual position - 𝑃𝑖𝑗 (i = 1 to m; j = 1 to n): The measured position reached by the
moving part on the jth
approach to the ith
target position.
Positional deviation -𝑥𝑖𝑗: Actual position reached by the moving part minus the target
position. It can be calculated according to (9).
𝑥𝑖𝑗 = 𝑃𝑖𝑗 − 𝑃𝑖 (9)
Unidirectional: The approach to a target position is always made in the same
direction. The symbol ↑ signifies an approach in the positive direction and ↓ signifies
an approach in the negative direction e.g. xij ↑ or xij ↓ .
Bi-directional: the approach to the target position is made in either direction.
Expanded uncertainty: the result of a measurement that is likely to encompass a
large portion of the distribution of values.
Coverage factor: This is a multiplier of the combined standard uncertainty in order
to obtain an expanded uncertainty.
Temperature: All measurements were carried out at room temperature 20º C. If this
was not the case, certain expansion coefficients may change. This could alter the
effectiveness of the results attained.
xix
Formulae
Mean unidirectional positional deviation at a position (𝑥𝑖 ↑ or 𝑥𝑖 ↓): Arithmetic
mean of the positional deviations obtained by a series of n unidirectional approaches
to a position Pi. They can be calculated according to (10) and (11).
𝑥𝑖 ↑=
1
𝑛∑ 𝑥𝑖𝑗
𝑛
𝑗=1
↑ (10)
𝑥𝑖 ↓=
1
𝑛∑ 𝑥𝑖𝑗
𝑛
𝑗=1
↓ (11)
Mean bi-directional positional deviation at a position (𝑥𝑖): Arithmetic mean of the
mean unidirectional positional deviations, 𝑥𝑖 ↑ and 𝑥𝑖 ↓ obtained from the two
directions of approach at a position Pi. It can be calculated according to (12).
𝑥𝑖 =
𝑥𝑖 ↑ +𝑥𝑖 ↓
2 (12)
Reversal value at a position (𝐵𝑖): Value of the difference between the mean
unidirectional positional deviations obtained from the two directions of approach at a
position 𝑃𝑖. It can be calculated according to (13).
𝐵𝑖 = 𝑥𝑖 ↑ −𝑥𝑖 ↓ (13)
xx
Reversal value of an axis (𝐵): Maximum of the absolute reversal values |𝐵𝑖| at all
target positions along or around the axis. It can be calculated according to (14).
𝐵 = 𝑚𝑎𝑥. [|𝐵𝑖|] (14)
Mean reversal value of an axis (𝐵): Arithmetic mean of the reversal values 𝐵𝑖 at all
target positions along or around the axis. It can be calculated according to (15).
𝐵 =
1
𝑚∑ 𝐵𝑖
𝑚
𝑖=1
(15)
Estimator of the unidirectional axis repeatability of positioning at a
position (𝑆𝑖 ↑ or 𝑆𝑖 ↓): Estimator of the standard uncertainty of the positional
deviations obtained by a series of n unidirectional approaches at a position 𝑃𝑖. They
can be calculated according to (16) and (17).
𝑠𝑖 ↑= √1
𝑛 − 1∑(𝑥𝑖𝑗 ↑ −𝑥𝑖 ↑)
2𝑛
𝑗=1
(16)
𝑠𝑖 ↓= √1
𝑛 − 1∑(𝑥𝑖𝑗 ↓ −𝑥𝑖 ↓)
2𝑛
𝑗=1
(17)
Unidirectional repeatability of positioning at a position 𝑅𝑖 ↑ or 𝑅𝑖 ↓ : Range
derived from the expanded uncertainty of unidirectional positional deviations at a
position 𝑃𝑖 using a coverage factor of 2. They can be calculated according to (18) and
(19).
𝑅𝑖 ↑= 4𝑠𝑖 ↑ (18)
𝑅𝑖 ↓= 4𝑠𝑖 ↓ (19)
xxi
Bi-directional repeatability of positioning at a position (𝑅𝑖): It can be calculated
according to (20).
𝑅𝑖 = 𝑚𝑎𝑥. [2𝑠𝑖 ↑ +2𝑠𝑖 ↓ +|𝐵𝑖|; 𝑅𝑖 ↑; 𝑅𝑖 ↓] (20)
Unidirectional repeatability of positioning (𝑅): Maximum value of the repeatability
of positioning at any position 𝑃𝑖 along or around the axis. It can be calculated
according to (21) and (22).
𝑅 ↑= 𝑚𝑎𝑥. [𝑅𝑖 ↑] (21)
𝑅 ↓= 𝑚𝑎𝑥. [𝑅𝑖 ↓] (22)
Bi-directional repeatability of positioning of an axis : Maximum value of the
repeatability of positioning at any position 𝑃𝑖 along or around the axis. It can be
calculated according to (23).
𝑅 = 𝑚𝑎𝑥. [𝑅𝑖] (23)
Unidirectional systematic positional deviation of an axis (𝐸 ↑ or 𝐸 ↓): The
difference between the algebraic maximum and minimum of the mean unidirectional
positional deviations for one approach direction 𝑥𝑖 ↑ or 𝑥𝑖 ↓ at any position 𝑃𝑖 along
or around the axis. They can be calculated according to (24) and (25).
𝐸 ↑= 𝑚𝑎𝑥. [𝑥𝑖 ↑] − 𝑚𝑖𝑛. [𝑥𝑖 ↑] (24)
𝐸 ↓= 𝑚𝑎𝑥. [𝑥𝑖 ↓] − 𝑚𝑖𝑛. [𝑥𝑖 ↓] (25)
xxii
Bi-directional systematic positional deviation of an axis (𝐸): The difference
between the algebraic maximum and minimum of the mean unidirectional positional
deviations for both approach direction 𝑥𝑖 ↑ or 𝑥𝑖 ↓ at any position 𝑃𝑖 along or around
the axis. It can be calculated according to (26).
𝐸 = 𝑚𝑎𝑥. [𝑥𝑖 ↑; 𝑥𝑖 ↓] − 𝑚𝑖𝑛. [𝑥𝑖 ↑; 𝑥𝑖 ↑] (26)
Range mean bi-directional positional deviation of an axis (𝑀): The difference
between the algebraic maximum and minimum of the mean bi-directional positional
deviations 𝑥𝑖 at any position 𝑃𝑖 along or around the axis. It can be calculated according
to (27).
𝑀 = 𝑚𝑎𝑥. [𝑥𝑖] − 𝑚𝑖𝑛. [𝑥𝑖] (27)
Unidirectional accuracy of positioning of an axis (𝐴 ↑ or 𝐴 ↓): Range derived from
the combination of the unidirectional systematic deviations and the estimator of the
standard uncertainty of unidirectional positioning using a coverage factor of 2. They
can be calculated according to (28) and (29).
𝐴 ↑= 𝑚𝑎𝑥. [𝑥𝑖 ↑ +2𝑠𝑖 ↑] − 𝑚𝑖𝑛. [𝑥𝑖 ↑ −2𝑠𝑖 ↑] (28)
𝐴 ↓= 𝑚𝑎𝑥. [𝑥𝑖 ↓ +2𝑠𝑖 ↓] − 𝑚𝑖𝑛. [𝑥𝑖 ↓ −2𝑠𝑖 ↓] (29)
Bi-directional accuracy of positioning of an axis (𝐴): Range derived from the
combination of the bidirectional systematic deviations and the estimator of the
standard uncertainty of bi-directional positioning using a coverage factor of 2. It can
be calculated according to (30).
𝐴 = 𝑚𝑎𝑥. [�̅�𝑖 ↑ +2𝑠𝑖 ↑; 𝑥𝑖 ↓ −2𝑠𝑖 ↓]
− 𝑚𝑖𝑛. [𝑥𝑖 ↑ −2𝑠𝑖 ↑; 𝑥𝑖 ↓ +2𝑠𝑖 ↓] (30)
xxiii
Appendix E : Fused Quartz Datasheet
xxiv
Appendix F
Channel Dimension Analysis
Table 44: Results from channel depth analysis
Sample /
Channel
Depth 1
(µm)
Depth 2
(µm)
Depth 3
(µm)
Depth
Average (µm)
Depth Std.
Dev. (µm)
1 545 572 572 563 16
2 297 294 303 298 5
3 157 160 154 157 3
4 661 668 694 674 17
5 608 584 620 604 19
6 781 804 781 789 14
7 769 769 807 782 22
8 703 668 703 691 20
9 670 657 670 666 8
10 262 257 252 257 5
11 122 121 117 120 3
12 74 73 75 74 1
13 638 638 638 638 0
14 498 488 513 500 13
15 375 379 368 374 6
16 974 993 955 974 19
17 852 878 878 869 15
18 748 763 763 758 9
19 155 152 149 152 3
20 95 100 98 98 2
21 72 75 71 72 2
22 462 439 480 460 21
23 364 368 375 369 6
24 286 289 272 282 9
25 757 734 795 762 31
26 551 557 573 560 11
27 569 569 592 577 13
xxv
Table 45: Results from channel width (top) analysis
Sample /
Channel
Width 1
(µm)
Width 2
(μm)
Width 3
(µm)
Width
Average (µm)
Width Std.
Dev. (µm)
1 213 207 204 208 4
2 176 181 174 177 4
3 172 177 170 173 4
4 197 195 197 196 1
5 182 173 182 179 5
6 206 214 196 205 9
7 193 183 185 187 5
8 170 162 170 167 5
9 170 167 165 167 3
10 179 175 175 177 2
11 167 175 165 169 5
12 149 145 149 148 3
13 185 187 179 184 4
14 179 186 184 183 4
15 185 178 183 182 4
16 191 193 191 192 1
17 188 182 186 185 3
18 209 209 203 207 4
19 185 189 178 184 6
20 173 180 176 176 3
21 146 147 140 145 4
22 191 197 191 193 3
23 197 199 197 198 1
24 185 191 185 187 3
25 206 200 196 201 5
26 200 210 190 200 10
27 178 185 169 177 8
xxvi
Table 46: Results from channel width (50%) analysis
Sample /
Channel
Width
50% 1
(µm)
Width
50% 2
(μm)
Width
50% 3
(µm)
Width 50%
Average (µm)
Width 50%
Std. Dev. (µm)
1 41 40 42 41 1
2 65 68 68 67 2
3 77 75 73 75 2
4 45 43 46 45 2
5 33 34 34 33 0
6 33 35 32 33 1
7 34 35 36 35 1
8 35 34 34 34 1
9 42 43 42 42 1
10 77 76 78 77 1
11 83 79 83 82 2
12 83 82 81 82 1
13 54 53 56 54 2
14 63 60 66 63 3
15 66 68 69 68 1
16 51 51 53 52 1
17 54 57 56 56 1
18 63 64 64 63 0
19 84 82 81 82 2
20 91 95 94 93 2
21 95 97 94 95 1
22 77 81 75 78 3
23 83 87 85 85 2
24 74 71 75 73 2
25 72 74 75 74 1
26 74 75 71 74 2
27 66 69 67 67 2
xxvii
Appendix G : 3D Analysis
Measurement results from the evaluation of the 3D measurement performance.
Run 1
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 5.4 516.2 510.8
Sub-run 2 6.5 516.9 510.4
Sub-run 3 5.1 517.2 512.1
Sub-run 4 4.7 516.9 512.2
Sub-run 5 4.3 517.2 512.9
Sub-run 6 3.9 515.2 511.3
Average Measured
Height (µm) 511.6
Step Height (µm) 512
Accuracy (%) 99.92
Standard Deviation
(µm) 0.9
Run 2
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 3.0 506.8 503.8
Sub-run 2 2.9 506.3 503.4
Sub-run 3 2.3 506.2 503.8
Sub-run 4 2.5 505.4 502.9
Sub-run 5 2.3 505.1 502.8
Sub-run 6 2.2 505.6 503.4
Average Measured
Height (µm) 503.4
Step Height (µm) 512
Accuracy (%) 98.32
Standard Deviation
(µm) 0.4
xxviii
Run 3
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 3.6 106.7 103.1
Sub-run 2 3.6 106.7 103.1
Sub-run 3 3.6 106.8 103.1
Sub-run 4 3.1 106.5 103.4
Sub-run 5 3.6 107.1 103.5
Sub-run 6 3.6 106.7 103.1
Average Measured
Height (µm) 103.2
Step Height (µm) 103
Accuracy (%) 99.81
Standard Deviation
(µm) 0.2
Run 4
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 3.1 106.2 103.1
Sub-run 2 3.4 106.5 103.1
Sub-run 3 3 106.3 103.2
Sub-run 4 3 106.2 103.2
Sub-run 5 3.3 106.2 102.9
Sub-run 6 3.5 106.6 103.1
Average Measured
Height (µm) 103.1
Step Height (µm) 103
Accuracy (%) 99.90
Standard Deviation
(µm) 0.1
xxix
Run 5
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 4.4 107 102.6
Sub-run 2 4.4 107.1 102.7
Sub-run 3 4.5 107.8 103.3
Sub-run 4 4.1 106.9 102.8
Sub-run 5 4.7 107.4 102.7
Sub-run 6 4.7 106.8 102.1
Average Measured
Height (µm) 102.7
Step Height (µm) 103
Accuracy (%) 99.71
Standard Deviation
(µm) 0.4
Run 6
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 3.4 105.8 102.4
Sub-run 2 3.5 106 102.5
Sub-run 3 3.6 106.1 102.5
Sub-run 4 3.5 106 102.4
Sub-run 5 3.6 106 102.4
Sub-run 6 3.5 105.9 102.4
Average Measured
Height (µm) 102.4
Step Height (µm) 103
Accuracy (%) 99.42
Standard Deviation
(µm) 0.1
xxx
Run 7
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 4.1 106.5 102.4
Sub-run 2 3.9 106.4 102.6
Sub-run 3 4.1 106.9 102.7
Sub-run 4 3.8 106.1 102.3
Sub-run 5 3.8 105.9 102.1
Sub-run 6 3.5 106.1 102.6
Average Measured
Height (µm) 102.5
Step Height (µm) 103
Accuracy (%) 99.51
Standard Deviation
(µm) 0.2
Run 8
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 2.5 506.7 504.2
Sub-run 2 2.8 506.4 503.6
Sub-run 3 2.9 506.6 503.7
Sub-run 4 2.8 505.8 503
Sub-run 5 3.3 505.9 502.7
Sub-run 6 2.8 505.3 502.5
Average Measured
Height (µm) 503.3
Step Height (µm) 512
Accuracy (%) 98.30
Standard Deviation
(µm) 0.7
xxxi
Run 9
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 15.6 118.7 103.1
Sub-run 2 15.7 118.7 103
Sub-run 3 15.5 118.3 102.8
Sub-run 4 15.6 118.7 103.1
Sub-run 5 15.7 118.5 102.7
Sub-run 6 15.8 118.7 102.9
Average Measured
Height (µm) 102.9
Step Height (µm) 103
Accuracy (%) 99.90
Standard Deviation
(µm) 0.1
Run 10
Lower Plane Height
(µm)
Upper Plane Height
(µm) Measured Height
(µm)
Sub-run 1 2.1 104.8 102.7
Sub-run 2 2.4 104.9 102.6
Sub-run 3 2.5 105.1 102.5
Sub-run 4 2.3 105 102.7
Sub-run 5 2.5 105 102.5
Sub-run 6 2.2 104.9 102.7
Average Measured
Height (µm) 102.6
Step Height (µm) 103
Accuracy (%) 99.61
Standard Deviation
(µm) 0.1